Wallbot

Settings reference / System & exchange

System & exchange settings

Exchanges and book sources, reconciliation, clock, shutdown, launcher and logs. 44 settings on this tab.

Generated from the code. Defaults, ranges, choices, units and the live confirmation rule on this page are read from the settings schema each time the docs are generated, so they always match the bot.

Summary

SettingInternal nameDefault
Order books frombook_sourcesbinance
Execution exchangeexecution_exchangebinance
Primary book sourceprimary_book_sourcebinance
Wall confirmationwall_confirmationprimary_only
Wall match tolerancewall_match_tolerance_bps5 bps
Max price-gap jumpmax_basis_jump_bps5 bps
Exchangeexchangesim
Paper account balancepaper_balance100 USDT
Paper USDC balancepaper_balance_usdc0 USDC
After a server restart: start pausedreboot_start_pausedOn
Price-gap windowbasis_window_s30 s
Maximum cross-exchange price gapmax_cross_gap_bps200 bps
Binance rate-limit budgetrate_limit_budget_pct70 %
Rejected key retrykey_retry_s600 s
Stop false-trigger windowfalse_trigger_window_s60 s
Safe shutdownsafe_shutdownOn
Keep a manual pause across restartspersist_manual_pauseOn
Retry failed cancels until confirmedcancel_retry_until_confirmedOn
Resync the clock on -1021clock_resync_on_1021On
Clock resync intervalclock_resync_min10 min
Periodic reconciliationperiodic_reconcileOn
Reconciliation intervalreconcile_interval_s60 s
Reconciliation grace for new ordersreconcile_grace_s15 s
Lost Wallbot orders are not foreign marginlost_orders_not_foreignOn
Refresh max leverage everyleverage_bracket_refresh_min60 min
No liquidation re-check while a position closesskip_liq_check_on_closeOn
Balance reads: cachebalance_cache_s3 s
Binance recvWindowrecv_window_ms10000 ms
Prepare to stop never aborts on exchange errorssafe_stop_error_tolerantOn
Take over a fill a cancel findsadopt_fill_on_cancelOn
Reconciliation adopts missed fillsreconcile_missed_fillsOn
Missed fill aftermissed_fill_after_s10 s
Route entry fills by order registryfill_route_by_registryOn
Reconnect: check pending entries' order statusreconnect_order_status_checkOn
Shutdown: close dashboards firstshutdown_close_dashboardsOn
Shutdown: connection timeoutshutdown_connection_timeout_s10 s
Launcher: log the shutdownlauncher_log_shutdownOn
Safe shutdown timeoutsafe_stop_timeout_s30 s
Launcher: wait before restartlauncher_restart_wait_s10 s
Launcher: restarts per hourlauncher_max_restarts_per_hour5
Launcher: clean stop on Linux signalslauncher_posix_signalsOn
Launcher: clean stop on Ctrl+Break (Windows)launcher_ctrl_breakOn
Log file: rotate atlog_rotate_mb50 MB
Log file: old files keptlog_keep5

All settings

Order books from

book_sources restarts the engine

Default
binance
Choices
binance, mexc (one or more)
Type
Multiple choice
Live confirmation
No. Changes apply in every mode without the typed confirmation.

In the dashboard: Which exchanges' order books the bot reads walls from (at least one). Only coins listed on every checked exchange and on the execution exchange are traded. Not used by the simulator.

What it does

Which exchanges' order books the bot reads walls from. You must tick at least one. Only coins listed on every ticked exchange and on the execution exchange can be traded. The offline simulator ignores this.

Example

Ticked: binance only. The bot looks for walls in the Binance book and trades them on Binance. Tick binance and mexc and a coin listed only on Binance drops out of the coin list, because it is not on both: if 40 coins qualify on Binance and 31 of them are also listed on MEXC, only those 31 remain.

When to change it

Leave it on binance for normal use. Add a second source only if you want to compare or confirm walls across exchanges (see wall_confirmation). MEXC support is built but has not been verified on this network.

Higher, lower, on or off

binance: walls come from the Binance USD-M futures book. mexc: walls come from the MEXC futures book. Both ticked: the bot reads both, and primary_book_source and wall_confirmation decide how they are combined.

Risk notes

Medium. Saving a change restarts the engine. Walls read on a different exchange than the one you trade on depend on the live price gap between the two (see max_basis_jump_bps).

Execution exchange

execution_exchange restarts the engine

Default
binance
Choices
binance, mexc
Type
Choice
Live confirmation
No. Changes apply in every mode without the typed confirmation.

In the dashboard: Where orders go (testnet / live) or are simulated (paper): leverage limits, fees, sizes, ticks, margin and balance all come from this exchange, and paper fills use its trades and mark/fair price. The Mode (Settings -> Account & keys) decides between paper, the exchange's test environment and the real exchange.

What it does

The exchange where orders are sent (testnet or live) or simulated (paper). Leverage limits, fees, order sizes, price ticks, margin and balance all come from this exchange, and paper fills use its trades and mark price. The Mode under Settings, Account decides paper, testnet or real.

Example

Execution binance, mode Paper: entries are simulated against Binance trades, with Binance tick sizes and fees (for example a BTCUSDT entry at 64250.0 on the 0.1 tick, with a 0.02% maker fee). Switch the mode to Testnet and the same orders go to the Binance futures testnet instead.

When to change it

Normally keep binance. Testnet mode only works with binance (MEXC has no futures test environment), and MEXC execution is built but not verified on this network.

Higher, lower, on or off

binance: orders, fees, sizes and balance from Binance USD-M futures. mexc: orders, fees, sizes and balance from MEXC futures (needs a saved, enabled MEXC key for live).

Risk notes

High. Saving a change restarts the engine. In live mode this is where real money is used, and the needed API key for that exchange must be saved and enabled or the bot refuses to start.

Primary book source

primary_book_source restarts the engine

Default
binance
Choices
binance, mexc
Type
Choice
Live confirmation
No. Changes apply in every mode without the typed confirmation.

In the dashboard: With two book sources: the exchange whose walls lead (see wall confirmation).

What it does

With two book sources ticked, the exchange whose walls lead. With wall_confirmation set to primary_only or both_required, new setups start only from walls on this exchange. With one book source it has no effect.

Example

Book sources binance and mexc, primary binance, confirmation both_required: a 2 million USDT bid wall on Binance is the candidate, and the bot then checks MEXC for a matching wall at the same price.

When to change it

Only when you use two book sources. Pick the exchange with the deeper, more reliable book for your coins.

Higher, lower, on or off

binance: Binance walls lead. mexc: MEXC walls lead. If the chosen one is not among the ticked book sources, the first ticked source is used instead.

Risk notes

Medium. Saving a change restarts the engine.

Wall confirmation

wall_confirmation

Default
primary_only
Choices
primary_only, both_required, either
Type
Choice
Live confirmation
No. Changes apply in every mode without the typed confirmation.

In the dashboard: With two book sources. Primary only: walls come from the primary source. Both required: the wall must also exist on the other exchange at the same price (after the price gap); a pull on either exchange then cancels or exits. Either: a wall on either exchange can trade.

What it does

With two book sources, how walls from the two exchanges are combined. It decides whether one exchange's wall is enough to trade, or whether both must show it. With one source, the bot always behaves as primary_only.

Example

both_required, tolerance 5 bps: a bid wall at 100.00 on Binance trades only if MEXC has a confirmed bid wall within 5 bps (100.00 plus or minus 0.05) after mapping prices. If MEXC's nearest is 100.08, the setup is skipped and logged.

When to change it

Try both_required if you see many spoofed walls on one exchange and want extra confirmation. Use either only if you want more setups and accept walls seen on just one exchange.

Higher, lower, on or off

primary_only: walls come only from primary_book_source. both_required: the wall must also exist on the other exchange at the same price (within wall_match_tolerance_bps); a pull on either exchange then cancels the entry or exits. either: a confirmed wall on either exchange can trade.

Risk notes

Medium. either allows more trades, including walls the other exchange does not see. both_required gives fewer trades and also reacts to pulls on the second exchange.

Wall match tolerance

wall_match_tolerance_bps

Default
5 bps
Range
0.1 to 100 bps
Type
Number
Unit
bps
Live confirmation
No. Changes apply in every mode without the typed confirmation.

In the dashboard: Both required: how close the two walls must be after mapping prices between exchanges.

What it does

Used only with wall_confirmation set to both_required. How close the two exchanges' walls must be, after mapping prices across the live price gap, to count as the same wall.

Example

Tolerance 5 bps on a coin at 100.00: the primary wall maps to 100.00 and the other exchange's wall maps to 100.04. That is 4 bps apart, so it matches. At 100.07 (7 bps) it would not match and the setup is skipped.

When to change it

Raise it if real walls are being rejected because the two books bucket prices slightly differently. Lower it if you want only near-exact matches.

Higher, lower, on or off

Higher: more walls count as matched, so more setups pass confirmation. Lower: stricter matching, fewer setups.

Risk notes

Low to medium. Too high can pair walls that are not really at the same level, which weakens the confirmation.

Max price-gap jump

max_basis_jump_bps

Default
5 bps
Range
0.1 to 200 bps
Type
Number
Unit
bps
Live confirmation
No. Changes apply in every mode without the typed confirmation.

In the dashboard: When walls come from a different exchange than the one executing: skip a setup if the price gap between the two moved more than this within the last 30 seconds.

What it does

When a wall comes from a different exchange than the one executing, the entry price is moved across by the live price gap between the two. If that gap moved more than this within the price-gap window (30 seconds by default), the setup is skipped.

Example

Walls from MEXC, orders on Binance, limit 5 bps. Over the last 30 seconds the MEXC/Binance gap went from 2 bps to 9 bps: a 7 bps jump, so the setup is skipped and the log says the gap moved 7.0 bps.

When to change it

Only matters with cross-exchange setups. Raise it if too many setups are skipped on volatile coins; lower it for stricter protection against a jumpy gap.

Higher, lower, on or off

Higher: more cross-exchange setups go ahead even when the gap is moving. Lower: setups are skipped more often while the gap is unstable.

Risk notes

Medium. A high value lets the bot place an entry translated through an unstable gap, so the order can land in the wrong spot relative to the real wall.

Exchange

exchange restarts the engine

Default
sim
Choices
sim, binance, mexc
Type
Choice
Live confirmation
No. Changes apply in every mode without the typed confirmation.

In the dashboard: Simulator (offline demo) or real markets. With Binance or MEXC selected, the book sources and the execution exchange above decide where data comes from and where orders go. 'sim' is an offline simulator for trying the dashboard.

What it does

The master switch between the offline simulator and real markets. With binance or mexc selected, book_sources and execution_exchange decide where data comes from and where orders go. sim is an offline demo for trying the dashboard.

Example

exchange = binance, book sources binance, execution binance, mode Paper: the bot reads the real Binance order book and simulates fills against the 100 USDT paper_balance. Set sim and it runs on made-up data with no network.

When to change it

Use sim only to explore the dashboard. Use binance for paper, testnet and live trading.

Higher, lower, on or off

sim: offline simulator, no real market data; testnet and live modes refuse to start with it. binance: real markets, with the book and execution settings above. mexc: real markets via MEXC (built, not verified on this network).

Risk notes

Medium. Saving a change restarts the engine.

Paper account balance

paper_balance

Default
100 USDT
Range
10 to 1e+06 USDT
Type
Number
Unit
USDT
Live confirmation
No. Changes apply in every mode without the typed confirmation.

In the dashboard: Starting balance for paper trading.

What it does

The starting balance for paper trading. After paper trades close (and when the bot stops), it remembers the running paper balance per execution exchange and continues from that, so this value is the starting point when no paper balance is saved yet.

Example

Paper balance 100 USDT on a fresh database: paper trading starts with 100 USDT. After a closed trade that lost 3 USDT, the saved balance is 97 USDT and the next start continues from 97, not 100.

When to change it

Set it before your first paper session to roughly match the money you plan to trade live, so sizes and minimum-order checks look realistic.

Higher, lower, on or off

Higher: bigger simulated margin per trade and fewer coins blocked by minimum order size. Lower: smaller trades, more coins too expensive to trade.

Risk notes

Low: paper only. A balance very different from your real one makes paper results less comparable to live.

Paper USDC balance

paper_balance_usdc

Default
0 USDC
Range
0 to 1e+06 USDC
Type
Number
Unit
USDC
Live confirmation
No. Changes apply in every mode without the typed confirmation.

In the dashboard: Starting balance of the paper USDC wallet (Contract type USDC or both). 0 = the same amount as the paper account balance.

What it does

The starting balance of the paper USDC wallet, used only when Contract type (contract_quote) is USDC or both. Like a Binance account in single-asset mode, paper keeps one wallet per asset: USDC contracts take their margin from the USDC wallet and their P&L and fees go back to it. 0 means the same amount as paper_balance. The running USDC balance is saved like the USDT one and continues at the next start.

Example

Contract type both, paper_balance 100 and paper_balance_usdc 0 on a fresh database: paper starts with 100 USDT and 100 USDC. A BTCUSDC trade with 2 USDC margin reserves 2 of the 100 USDC; the USDT wallet is untouched.

When to change it

Before a paper test of USDC contracts, set it to the USDC you would move into the futures wallet for live trading.

Higher, lower, on or off

Higher: bigger simulated USDC margin and fewer USDC coins blocked by the minimum order size. Lower: smaller USDC trades. No effect with Contract type USDT.

Risk notes

Low: paper only.

After a server restart: start paused

reboot_start_paused

Default
On
Range
On or off
Type
On/off
Live confirmation
No. Changes apply in every mode without the typed confirmation.

In the dashboard: When the @reboot crontab entry starts the Trading instance after an unplanned server restart, new entries start PAUSED (only Resume lifts it) and you get a high-priority notification.

What it does

Trading instance only: when it is started by the @reboot crontab entry (an unplanned server restart), new entries start PAUSED as if you had pressed Pause; only Resume entries lifts it. A normal start or a launcher crash restart is not affected.

Example

On: the VPS restarts at 03:12; at 03:14 the Trading instance starts, reconciles the 1 open position with its stop, opens nothing new and notifies you; you check and press Resume at 07:30.

When to change it

Leave it on. Turn it off only if you want trading to continue by itself after a restart.

Higher, lower, on or off

On: nothing new is opened until you have looked. Off: entries continue right after the restart.

Risk notes

Low when on (missed entries only). Off: trading resumes without anyone checking the state after the restart.

Price-gap window

basis_window_s

Default
30 s
Range
5 to 600 s
Type
Number
Unit
s
Live confirmation
No. Changes apply in every mode without the typed confirmation.

In the dashboard: The book-source / execution price gap is averaged, and its jump measured, over this window.

What it does

When walls come from a different exchange than the one you trade on, the bot measures the price gap between the two. The gap is averaged over this window, and the jump (highest minus lowest gap) is also measured over it and compared with max_basis_jump_bps.

Example

Window 30 s: over the last 30 s the gap between the book source and Binance ranged from +2 to +9 bps, a jump of 7 bps. With max_basis_jump_bps at 5, the wall is skipped with "price gap moved 7.0 bps in the last 30 s".

When to change it

Only matters when you use a book source other than the execution exchange. Longer smooths the gap; shorter reacts faster.

Higher, lower, on or off

Higher: a smoother average gap, but a wider window usually shows a bigger jump, so more walls are skipped. Lower: the average follows the latest gap, jumps look smaller, fewer skips. New windows take effect when the gap trackers are rebuilt (for example after a restart).

Risk notes

Low to medium. A too short window can let entries through while two markets are briefly out of line.

Maximum cross-exchange price gap

max_cross_gap_bps

Default
200 bps
Range
10 to 5000 bps
Type
Number
Unit
bps
Live confirmation
No. Changes apply in every mode without the typed confirmation.

In the dashboard: A book source whose price differs from the execution exchange by more than this is treated as a different contract and skipped.

What it does

Safety check for cross-exchange book sources. If the average price gap between the book source and the execution exchange is larger than this, the bot assumes they are not the same contract (for example a renamed or rebased coin) and skips the wall with a warning.

Example

Limit 200 bps (2%): a coin trades at 1.000 on the source and 1.030 on Binance, a 300 bps gap. The wall is skipped with "prices differ by 300 bps (different contract?)".

When to change it

Only matters with a book source other than the execution exchange. Lower it to be stricter; raise it only if you know a coin legitimately trades at a larger gap.

Higher, lower, on or off

Higher: more cross-exchange walls are accepted, including possibly mismatched contracts. Lower: walls are skipped on smaller gaps, safer but fewer entries.

Risk notes

High if raised a lot: entries could be placed using walls from a market that is not the same instrument.

Binance rate-limit budget

rate_limit_budget_pct

Default
70 %
Range
10 to 95 %
Type
Number
Unit
%
Live confirmation
No. Changes apply in every mode without the typed confirmation.

In the dashboard: Non-critical requests wait when used weight or order counts reach this share of Binance's published limits. Cancels, stops, closes and margin top-ups never wait.

What it does

The bot keeps its own count of Binance request weight and order counts. Non-critical requests (scans, candles, new entries and similar) wait when usage reaches this share of Binance's published limits. Cancels, stops, closes and margin top-ups never wait and go out on a separate connection.

Example

Budget 70% with a request-weight limit of 2,400 per minute: non-critical requests pause once about 1,680 weight has been used in the current minute, leaving 720 for protective orders.

When to change it

Lower it if you share the API key or IP with other tools, or if you see 429 warnings. Raise it only if scans or entries are often delayed and nothing else uses the key.

Higher, lower, on or off

Higher: more of the limit available for routine work, less headroom for protective orders and other tools. Lower: more headroom, routine work waits more often.

Risk notes

Medium to high if raised: hitting Binance's real limit causes 429 back-offs or a temporary IP ban (418), and new entries pause.

Rejected key retry

key_retry_s

Default
600 s
Range
60 to 86400 s
Type
Whole number
Unit
s
Live confirmation
No. Changes apply in every mode without the typed confirmation.

In the dashboard: Paper mode: while the saved read-only key is rejected, it is retried this often.

What it does

Paper mode only. If Binance rejects the saved API key (for example the IP is not whitelisted), paper trading continues with cached or assumed leverage brackets and fees. The bot retries the key this often and switches back by itself when it works.

Example

600 s: the key is rejected at 08:00. The bot tries again at 08:10, 08:20 and so on. If you whitelist the IP at 08:14, the key is accepted at 08:20 and live brackets and fees are used again.

When to change it

Lower it if you are fixing the key and want it picked up sooner. Raise it to reduce retries while the key is known to be broken.

Higher, lower, on or off

Higher: fewer retries, slower recovery after a fix. Lower: quicker recovery, more rejected requests.

Risk notes

Low. Live mode does not use this; it refuses to trade with a rejected key.

Stop false-trigger window

false_trigger_window_s

Default
60 s
Range
5 to 3600 s
Type
Number
Unit
s
Live confirmation
No. Changes apply in every mode without the typed confirmation.

In the dashboard: Live learning: a stop counts as a false trigger when price comes back past it within this time.

What it does

Live learning log only (testnet and live). After a stop fills, the bot watches the price for this long. If price comes back to the safe side of the stop within the window, the stop is logged as a "false trigger". Nothing is traded on this.

Example

60 s: a long's stop at 99.50 fills at 12:00:00. Price is back at 99.60 at 12:00:40, so a stop_false_trigger event is logged with 40 seconds after the stop. Had it recovered at 12:01:10, nothing would be logged.

When to change it

Lengthen it to count slower recoveries as false triggers in your analysis. Shorten it to count only quick wicks.

Higher, lower, on or off

Higher: more stops are labelled false triggers. Lower: only fast wicks count.

Risk notes

Low. Only changes what is logged.

Safe shutdown

safe_shutdown

Default
On
Range
On or off
Type
On/off
Live confirmation
No. Changes apply in every mode without the typed confirmation.

In the dashboard: On shutdown in testnet and live: stop new entries, wait for order actions in progress, cancel and confirm every resting entry, keep open positions with their exchange stops, then exit. If that cannot complete, a critical alert is sent before exiting.

What it does

Testnet and live: when the bot shuts down it first stops new entries, waits for order actions in progress, cancels every resting entry and confirms each cancel, and leaves open positions with their exchange stops. If this cannot be completed, it logs "SHUTTING DOWN WHILE NOT SAFE", sends a critical alert, and the next start reconciles.

Example

2 entries are resting and 1 position is open when you stop the bot. With this on, both entries are cancelled and confirmed, the position keeps its stop on Binance, then the bot exits.

When to change it

Keep it on. Turning it off only makes sense for quick restarts where you will check the exchange yourself.

Higher, lower, on or off

On: resting entries are cleaned up before exit; shutdown can take up to the safe-stop timeout. Off: the bot exits immediately and resting entries stay on the exchange until the next start cancels them.

Risk notes

High if off: entries left resting can fill while the bot is not running, giving a position the bot only finds at the next start.

Keep a manual pause across restarts

persist_manual_pause

Default
On
Range
On or off
Type
On/off
Live confirmation
No. Changes apply in every mode without the typed confirmation.

In the dashboard: Pause entries and Prepare to stop survive a restart: the bot starts paused, says so on the dashboard and in a notification, and only Resume entries lifts it. The daily-loss pause still lifts at the next UTC day.

What it does

"Pause entries" and "Prepare to stop" are saved and survive a restart. The bot then starts paused, says so on the dashboard and in a notification, and only "Resume entries" lifts it. The daily-loss pause is separate and still lifts at the next UTC day.

Example

You click "Prepare to stop" at 18:00, restart the bot at 18:05 after an update. With this on, it starts paused and places no entries until you click "Resume entries".

When to change it

Keep it on, especially on a live instance. Turn it off only if you want every restart to start trading straight away.

Higher, lower, on or off

On: a manual pause outlives restarts. Off: the pause works for the current session only; after a restart the bot trades again.

Risk notes

Medium if off: a restart (including a crash and auto-restart) can resume live entries you meant to keep paused.

Retry failed cancels until confirmed

cancel_retry_until_confirmed

Default
On
Range
On or off
Type
On/off
Live confirmation
No. Changes apply in every mode without the typed confirmation.

In the dashboard: Testnet and live: an entry cancel the exchange does not confirm is retried (1, 2, 5, 10 s, then every 30 s) until Binance confirms the order is cancelled, filled or absent. While this is on, the trade is never marked cancelled on an unconfirmed cancel. Off: the trade is marked cancelled at once and the order goes on the unconfirmed-cancels list, which the reconciliation keeps retrying.

What it does

Testnet and live: if Binance does not confirm an entry cancel, the bot retries after 1, 2, 5 and 10 s, then every 30 s, until Binance confirms the order is cancelled, filled or absent. While this is on, the trade stays tracked as pending meanwhile and is never marked cancelled on an unconfirmed cancel.

Example

A wall is pulled and the bot cancels its ETH entry, but Binance times out. The bot retries at +1 s, +3 s, +8 s, +18 s, then every 30 s. If the retry finds the order filled, the fill is taken over and protected.

When to change it

Keep it on, especially in live mode.

Higher, lower, on or off

On: unconfirmed cancels are chased and the trade stays tracked; cancel_alert_after_s sends an alert if it drags on. Off: a critical log line is written, the order is listed as unconfirmed, the trade is marked cancelled locally, and the cancel is only retried by "Prepare to stop" or at the next start.

Risk notes

High if off: an order that is still resting can fill without the bot tracking it.

Resync the clock on -1021

clock_resync_on_1021

Default
On
Range
On or off
Type
On/off
Live confirmation
No. Changes apply in every mode without the typed confirmation.

In the dashboard: When Binance rejects a request with -1021 (timestamp outside recvWindow), resync with Binance's server time and retry that request once. A -1021 request was never executed, so the retry is safe.

What it does

Binance rejects a signed request with error -1021 when its timestamp falls outside the allowed window (the PC clock has drifted). With this on, the bot resyncs with Binance's server time and retries that request once. A -1021 request was never executed, so the retry cannot double an order.

Example

The PC clock drifts 1.5 s ahead and a stop placement gets -1021. The bot fetches Binance's time, corrects its offset (for example -1,500 ms) and sends the stop again straight away.

When to change it

Keep it on. Turn it off only to diagnose clock problems.

Higher, lower, on or off

On: drift errors are fixed and the request is retried once automatically. Off: the request fails and the bot handles it as a normal error (an entry is skipped, a protective order is retried by its own logic later).

Risk notes

Medium if off: a protective order can be delayed by a clock drift error.

Clock resync interval

clock_resync_min

Default
10 min
Range
1 to 1440 min
Type
Number
Unit
min
Live confirmation
No. Changes apply in every mode without the typed confirmation.

In the dashboard: Resync with Binance's server time this often and log the offset (testnet and live).

What it does

Testnet and live: the bot resyncs its clock with Binance's server time this often and logs the offset, so signed requests keep valid timestamps.

Example

10 min: the clock is synced at 09:00, 09:10, 09:20 and so on. If the PC drifts 200 ms in an hour, each sync corrects the 30 to 40 ms gained since the last one.

When to change it

Lower it if your PC clock drifts quickly and you see -1021 errors. Raise it to save a few requests.

Higher, lower, on or off

Higher: fewer sync requests, more drift can build up between syncs. Lower: more frequent syncs, timestamps stay closer to Binance's.

Risk notes

Low. clock_resync_on_1021 catches drift between syncs.

Periodic reconciliation

periodic_reconcile

Default
On
Range
On or off
Type
On/off
Live confirmation
No. Changes apply in every mode without the typed confirmation.

In the dashboard: Testnet and live: compare Binance's open orders, algo orders and positions with the bot's trades every interval and after any exchange error. Lost Wallbot orders (our client-id prefix, not tracked) are cancelled; a lost Wallbot position is adopted with its planned stop if still valid, else closed at market, always with a high-priority notification. Orders and positions without our prefix are yours and are never touched.

What it does

On testnet and live, Wallbot regularly compares Binance's open orders, stop orders and positions with its own trade list. Lost Wallbot orders (carrying the bot's client-id prefix but not tracked) are cancelled. A lost Wallbot position is adopted with its planned stop if that stop is still valid, otherwise closed at market, with a high-priority notification. Orders and positions without the prefix are yours and are never touched.

Example

With reconcile_interval_s at 60, a pass runs every 60 seconds. If an exchange error happens at 12:00:20, another pass runs about 10 seconds after the previous one instead of waiting the full minute. A Wallbot entry order found resting on ETHUSDT that the bot no longer tracks is cancelled in that pass.

When to change it

Leave it on. Turning it off only makes sense for a short debugging session on testnet.

Higher, lower, on or off

On: lost orders and positions are found and handled within about a minute, and missed fills and unprotected positions are detected (reconcile_missed_fills, unprotected_alert run inside this pass). Off: none of these checks run; a lost order can sit on the exchange holding margin, or fill into a position with no stop, until the next restart.

Risk notes

High if switched off: this is the safety net for orders the bot lost track of (it was added after real cases of untracked orders on the live account). Paper mode is not affected.

Reconciliation interval

reconcile_interval_s

Default
60 s
Range
15 to 3600 s
Type
Number
Unit
s
Live confirmation
No. Changes apply in every mode without the typed confirmation.

In the dashboard: How often the periodic reconciliation runs.

What it does

How often the periodic reconciliation (periodic_reconcile) runs on testnet and live. An exchange error also triggers an extra pass, at most once every 10 seconds.

Example

At 60 seconds, a lost Wallbot order is found and cancelled within about one minute. At 300 seconds it could rest on the exchange for up to five minutes first.

When to change it

Lower it (for example 30) if you want faster detection and have plenty of API weight to spare. Raise it only if Binance rate limits become a problem.

Higher, lower, on or off

Higher: fewer API calls, but lost orders, missed fills and unprotected positions are found later. Lower: faster detection, more API calls (each pass lists all open orders, stop orders and positions).

Risk notes

Medium: a long interval delays every safety check that lives inside the reconciliation, including the unprotected-position alert.

Reconciliation grace for new orders

reconcile_grace_s

Default
15 s
Range
5 to 300 s
Type
Number
Unit
s
Live confirmation
No. Changes apply in every mode without the typed confirmation.

In the dashboard: Orders younger than this are left alone (the bot may not have recorded them yet).

What it does

Orders and stop orders younger than this are skipped by the reconciliation, because the bot may have just placed them and not yet recorded their ids.

Example

At 15 seconds: an entry placed at 12:00:00 that the reconciliation sees at 12:00:08 is left alone. If it is still untracked at the pass after 12:00:15, it is treated as lost and cancelled.

When to change it

Raise it if you ever see the bot cancel its own brand-new orders as "lost" (for example on a slow connection). There is rarely a reason to lower it.

Higher, lower, on or off

Higher: safer for new orders, but a truly lost order rests a little longer. Lower: lost orders are cancelled sooner, with a higher chance of cancelling an order the bot is still recording.

Risk notes

Medium if set too low: the bot could cancel a fresh entry or stop it has just placed.

Lost Wallbot orders are not foreign margin

lost_orders_not_foreign

Default
On
Range
On or off
Type
On/off
Live confirmation
No. Changes apply in every mode without the typed confirmation.

In the dashboard: Margin held by lost Wallbot orders is shown as such (and handled by the reconciliation), not as margin used by other positions or orders; new entries on that symbol wait until it is cancelled.

What it does

Margin held by lost Wallbot orders whose cancel has not been confirmed is shown as "held by lost Wallbot order(s)" in its own warning, and is not counted as foreign margin (margin used by positions or orders Wallbot did not open).

Example

Balance 100 USDT, available 80, Wallbot's tracked trades use 12, and a lost Wallbot order on ETHUSDT holds 8 that could not be cancelled yet. On: foreign margin is 100 - 80 - 12 - 8 = 0, and a warning says 8.00 USDT is held by lost orders being cancelled. Off: the banner reports 8.00 USDT of foreign margin.

When to change it

Leave it on. It only changes how the margin is labelled.

Higher, lower, on or off

On: lost-order margin gets its own clear warning, separate from foreign margin. Off: that margin is mixed into the foreign-margin figure, which can make you think another app or a manual trade is using your account. In both cases new entries on that coin wait until the lost order is cancelled.

Risk notes

Low: display and warning text only; the reconciliation still cancels the lost order either way.

Refresh max leverage every

leverage_bracket_refresh_min

Default
60 min
Range
5 to 1440 min
Type
Whole number
Unit
min
Live confirmation
No. Changes apply in every mode without the typed confirmation.

In the dashboard: Binance's per-coin maximum leverage (this account's leverage brackets) is re-read this often, and at once after a leverage mismatch or a leverage-related rejection. A changed maximum is logged and notified.

What it does

How often Wallbot re-reads this account's leverage brackets on Binance (each coin's maximum leverage and maintenance tiers). It is also re-read at once after a leverage mismatch, a leverage-related rejection or a "max leverage" error. When a watched or traded coin's maximum changes, the log and a notification say e.g. "LTCUSDT max leverage on Binance changed: 75x → 10x".

Example

At 60 the brackets are at most an hour old. If Binance lowers LTCUSDT from 75x to 20x at 14:10, the next refresh (by 15:10, or at once on a rejection) plans every new LTC entry at 20x or less.

When to change it

Lower it if Binance changes leverage tiers often for the coins you trade.

Higher, lower, on or off

Lower: fresher maximums, a few more requests (one per coin per refresh, weight 1). Higher: fewer requests.

Risk notes

Low. Planning with an out-of-date maximum gets the entry rejected or capped, which now also triggers a refresh.

No liquidation re-check while a position closes

skip_liq_check_on_close

Default
On
Range
On or off
Type
On/off
Live confirmation
No. Changes apply in every mode without the typed confirmation.

In the dashboard: Binance sends a position update for every piece of a closing fill. While the position is 0 or smaller than the trade (a stop, take profit or close executing), Wallbot skips the liquidation re-check for it, so the close is recorded at once instead of after several extra requests.

What it does

A closing fill (stop, take profit, close) often arrives in several pieces, and Binance sends a position update for each. With this on, Wallbot skips the liquidation re-check while the position is 0 or smaller than the trade, so the close itself is recorded right away.

Example

ARBUSDT: the stop's market order filled 786.5 in 7 pieces at 14:02:06. Off: 6 position reads (about 0.4 to 1.3 s each) ran before the close was recorded, about 7 s later. On: the close is recorded straight after the last piece.

When to change it

Leave it on. Turn it off only to re-check liquidation after every partial fill.

Higher, lower, on or off

On: faster close records, fewer requests. Off: a liquidation re-check per position update, even while the position is being closed.

Risk notes

None: reducing an isolated position never brings its liquidation closer, and the full check still runs whenever the position is unchanged or larger.

Balance reads: cache

balance_cache_s I ACCEPT in live

Default
3 s
Range
0 to 30 s
Type
Number
Unit
s
Live confirmation
Yes: any change in live mode asks you to type I ACCEPT.

In the dashboard: Binance wallet balance reads are served from a cache at most this old. Any order placed, changed, cancelled or refused and every user-data event (fills, order updates, balance and position updates) clears it, so the next read goes to Binance; with the user stream down every read does. 0 = every read goes to Binance.

What it does

Wallet balance reads (sizing and the free-balance check before each entry, the dashboard, the 30-second refresh) are served from a cache at most this many seconds old. Any order Wallbot places, changes or cancels, any refusal, and every user-data event (fills, order updates, balance and position updates) clears the cache, so the next read goes to Binance. With the user stream disconnected every read goes to Binance. Concurrent reads share one request. Efficiency only: it changes no trading decision.

Example

3: ten wall touches within 3 s with nothing changing on the account: one balance request instead of twenty.

When to change it

Always; lower it only to debug.

Higher, lower, on or off

Higher: fewer requests; 0: every read is a request (as before r40).

Risk notes

Very low: a change Binance does not report by an event (for example unrealized PnL of a cross position) shows up at most this late.

Binance recvWindow

recv_window_ms

Default
10000 ms
Range
1000 to 60000 ms
Type
Whole number
Unit
ms
Live confirmation
No. Changes apply in every mode without the typed confirmation.

In the dashboard: How late a signed request may reach Binance before it is rejected with -1021. A larger window rides out slow, jittery connections; Binance allows up to 60000 ms.

What it does

Binance rejects a signed request with -1021 when it arrives more than this long after its timestamp. A larger window keeps requests valid on a slow connection.

Example

Default 10000: a cancel that takes 6.0 s to reach Binance on a congested line is still accepted; with 5000 it would be rejected and retried.

When to change it

Raise it (up to 60000) on a slow or unstable connection; keep it lower on a fast, stable one.

Higher, lower, on or off

Higher: fewer -1021 rejections during slow periods. Lower: stricter; stale requests are refused sooner.

Risk notes

Low: a request that arrives late can still execute, which matters little for the order types Wallbot uses.

Prepare to stop never aborts on exchange errors

safe_stop_error_tolerant

Default
On
Range
On or off
Type
On/off
Live confirmation
No. Changes apply in every mode without the typed confirmation.

In the dashboard: A failed position check after a failed cancel counts as unknown and the cancel is retried until confirmed; Prepare to stop and shutdown never abort on an exchange error and always finish with an honest Safe or NOT SAFE (NOT SAFE is notified).

What it does

Makes "Prepare to stop" and shutdown robust against exchange errors. If a cancel fails and the follow-up position check also fails, the order counts as unknown and its cancel keeps being retried until confirmed. Prepare to stop and shutdown then always finish with an honest "Safe to stop: yes" or "NOT SAFE" (NOT SAFE is notified).

Example

You press "Prepare to stop" with two pending entries. The first cancel gets a timeout and the position check gets -1021. On: that entry is chased as unconfirmed, the second entry is still cancelled, and the result is NOT SAFE with the order id and error. Off: the error aborts the preparation and the second entry may stay resting.

When to change it

Leave it on. Off only reproduces the old behaviour for testing.

Higher, lower, on or off

On: one exchange error never stops the safe-stop routine part way; problems are listed in the NOT SAFE reason. Off: the first exchange error raises and aborts Prepare to stop, restart retries or shutdown, possibly leaving orders on the exchange with no clear report.

Risk notes

High if off: a stop that aborts mid-way can leave a resting entry that later fills with nobody watching it.

Take over a fill a cancel finds

adopt_fill_on_cancel

Default
On
Range
On or off
Type
On/off
Live confirmation
No. Changes apply in every mode without the typed confirmation.

In the dashboard: When cancelling an entry finds it already filled, Wallbot applies the fill itself at once (de-duplicated with the fill event) and places the stop, instead of waiting for a fill event.

What it does

When Wallbot tries to cancel an entry and finds it already filled, it applies the fill itself straight away (de-duplicated with the normal fill event) and places the stop. It does not wait for a fill event that might never arrive. Testnet and live only.

Example

A spoof is detected and the bot cancels its 0.003 BTC entry at 64250.0, but the cancel fails because the order just filled. On: the log says "Cancel raced a fill ... taking the fill over now" and the stop is placed within moments. Off: the bot waits for the fill event; if that event was missed, the position has no stop.

When to change it

Leave it on.

Higher, lower, on or off

On: a fill discovered by a cancel is adopted at once with its stop. Off: the bot relies on the fill event, or reports the fill as lost when a cancel retry finds it.

Risk notes

High if off: a missed fill event leaves a real position without a stop (this happened on the live account and led to this switch).

Reconciliation adopts missed fills

reconcile_missed_fills

Default
On
Range
On or off
Type
On/off
Live confirmation
No. Changes apply in every mode without the typed confirmation.

In the dashboard: Testnet and live: a real position with a matching trade Wallbot still treats as pending is a missed fill: after missed_fill_after_s it is taken over with its stop, without a restart.

What it does

On testnet and live, the reconciliation looks for a real position on Binance where Wallbot still has the matching trade as pending. That is a missed fill. Once it has lasted missed_fill_after_s, the position is taken over with its stop, without a restart.

Example

An ETHUSDT entry fills but the fill message never arrives. The next reconciliation pass sees a long ETHUSDT position while the trade is still pending. It re-checks about 10 seconds later, then adopts the fill and places the planned stop.

When to change it

Leave it on. It needs periodic_reconcile on.

Higher, lower, on or off

On: missed fills are repaired within roughly one reconciliation interval. Off: the position stays unmanaged (no stop from the bot) until a restart or a cancel discovers it; unprotected_alert still warns you.

Risk notes

High if off: an unprotected position can stay open for hours (a real live case lasted 6 hours).

Missed fill after

missed_fill_after_s

Default
10 s
Range
3 to 300 s
Type
Number
Unit
s
Live confirmation
No. Changes apply in every mode without the typed confirmation.

In the dashboard: How long a position with a pending trade may last before the reconciliation takes it over.

What it does

How long a real position may exist while its Wallbot trade is still pending before the reconciliation takes it over as a missed fill. The wait gives the normal fill event a chance to arrive first.

Example

At 10 seconds: the reconciliation first sees the position at 12:00:00 and checks again about 10 seconds later. If the fill has still not been applied, it adopts the fill and places the stop at that point.

When to change it

Lower it slightly for faster protection; raise it if your connection often delivers fill events late.

Higher, lower, on or off

Higher: fewer chances of racing the normal fill event, but the position stays without a stop for longer. Lower: faster takeover, with the adoption de-duplicated against a late fill event.

Risk notes

Medium if set high: the position has no stop for this long, plus the time until the next pass.

Route entry fills by order registry

fill_route_by_registry

Default
On
Range
On or off
Type
On/off
Live confirmation
No. Changes apply in every mode without the typed confirmation.

In the dashboard: Entry fills are recognised by the bot's own order registry, not by the client-id prefix (the prefix test mistook about 1 in 16 entries, ids 'wbc...', for close orders and dropped their fills).

What it does

Entry fills from Binance are recognised by the bot's own list of take-profit order ids, not by the client-id prefix. The old prefix test mistook about 1 in 16 entries (random ids starting "wbc") for close orders and dropped their fills.

Example

An entry order gets the client id "wbc4f...". On: the bot sees it is not one of its take-profit orders and applies the fill, placing the stop. Off: the prefix test treats "wbc" as a close order and ignores the fill, leaving the position without a stop.

When to change it

Leave it on. Off exists only to reproduce the old bug.

Higher, lower, on or off

On: every entry fill is applied. Off: roughly 1 in 16 entry fills is silently dropped.

Risk notes

High if off: this exact bug left a live position unprotected for hours.

Reconnect: check pending entries' order status

reconnect_order_status_check

Default
On
Range
On or off
Type
On/off
Live confirmation
No. Changes apply in every mode without the typed confirmation.

In the dashboard: On every user-data reconnect or key renewal, each pending entry is checked against Binance's order status: a filled one is applied at once, a cancelled one is closed.

What it does

On every user-data reconnect or key renewal (testnet and live), each pending entry is checked against Binance's order status. A filled one is applied at once; a cancelled one is closed in the bot.

Example

The user-data stream drops for 20 seconds and an entry fills during the gap. On reconnect, Binance reports the order as filled: the log says "Reconnect: entry ... is filled on Binance; applying the fill" and the stop is placed.

When to change it

Leave it on.

Higher, lower, on or off

On: fills and cancels missed during a disconnect are caught at reconnect. Off: they are only caught later by the reconciliation (reconcile_missed_fills) or a restart.

Risk notes

Medium if off: a fill during a disconnect can wait longer for its stop.

Shutdown: close dashboards first

shutdown_close_dashboards

Default
On
Range
On or off
Type
On/off
Live confirmation
No. Changes apply in every mode without the typed confirmation.

In the dashboard: On stop, close every open dashboard connection first, then run the safe shutdown, then exit, so an open browser tab never blocks it and the clean-stop flag is always recorded (read at start).

What it does

On stop, Wallbot first closes every open dashboard connection, then runs the safe shutdown, then exits. An open browser tab can then never block the shutdown, and the clean-stop flag is always recorded. Read when the bot starts.

Example

You press Ctrl+C with the dashboard open in 2 tabs. On: both tabs disconnect at once and the safe shutdown runs. Off: the web server waits for the tabs to close, which they never do by themselves, so the safe shutdown may not run (on: open connections get at most shutdown_connection_timeout_s, 10 seconds).

When to change it

Leave it on. A change takes effect at the next start.

Higher, lower, on or off

On: shutdown never waits on dashboard tabs, and shutdown_connection_timeout_s acts as a backstop. Off: no connection timeout either, so shutdown can hang on an open tab.

Risk notes

Medium if off: a hung shutdown can skip the safe stop and the clean-stop flag.

Shutdown: connection timeout

shutdown_connection_timeout_s

Default
10 s
Range
1 to 120 s
Type
Number
Unit
s
Live confirmation
No. Changes apply in every mode without the typed confirmation.

In the dashboard: Backstop: other connections still open this long after the stop request are cut so the safe shutdown runs (read at start).

What it does

Backstop during shutdown: connections still open this long after the stop request (for example a request in progress) are cut so the safe shutdown can run. Read when the bot starts, and only used when shutdown_close_dashboards is on.

Example

At 10 seconds: a slow dashboard request is still running 10 seconds after Ctrl+C, so it is cut and the safe shutdown begins. The launcher waits up to 10 + safe_stop_timeout_s (30) + 35 = 75 seconds for the bot to finish.

When to change it

Raise it only if long requests are being cut during shutdown. A change takes effect at the next start.

Higher, lower, on or off

Higher: in-progress requests get more time, shutdown can take longer. Lower: faster shutdown, requests cut sooner.

Risk notes

Low: affects shutdown speed, not trading.

Launcher: log the shutdown

launcher_log_shutdown needs restart

Default
On
Range
On or off
Type
On/off
Live confirmation
No. Changes apply in every mode without the typed confirmation.

In the dashboard: After Ctrl+C the launcher keeps writing the bot's output to the log file until the bot has exited, so the safe-stop lines are kept (read at launcher start).

What it does

After Ctrl+C, the launcher (start_wallbot.py) keeps writing the bot's output to the log file until the bot has exited, so the safe-stop lines are kept. Read when the launcher starts.

Example

You press Ctrl+C. On: the lines "Preparing to stop ..." and "Safe to stop: ..." end up in the log file. The launcher waits up to about 75 seconds with the default timeouts. Off: the launcher stops logging at once and those lines are lost.

When to change it

Leave it on. A change takes effect the next time you start the launcher.

Higher, lower, on or off

On: the full shutdown is logged. Off: the log ends at Ctrl+C, so you cannot check afterwards whether the stop was safe.

Risk notes

Low for trading, but off removes the record you need to confirm a safe stop.

Safe shutdown timeout

safe_stop_timeout_s

Default
30 s
Range
5 to 600 s
Type
Number
Unit
s
Live confirmation
No. Changes apply in every mode without the typed confirmation.

In the dashboard: How long 'Prepare to stop' and shutdown wait for order actions in progress.

What it does

How long "Prepare to stop" and shutdown wait for order actions in progress (placing, cancelling, moving stops) before reporting. Actions still running after this are listed in the NOT SAFE reason. The launcher also uses it to decide how long to wait for the bot to exit.

Example

At 30 seconds: you press "Prepare to stop" while a stop is being moved. If it finishes within 30 seconds, the result can be "Safe to stop: yes". If not, you get NOT SAFE with "order actions still running after 30 s".

When to change it

Raise it if Binance is slow and you often see NOT SAFE only because of the timeout.

Higher, lower, on or off

Higher: more time for in-progress orders to finish, slower stop. Lower: faster answer, more likely NOT SAFE.

Risk notes

Medium if too low: you may be told NOT SAFE more often; never stop a live bot until it says Safe to stop.

Launcher: wait before restart

launcher_restart_wait_s needs restart

Default
10 s
Range
1 to 600 s
Type
Number
Unit
s
Live confirmation
No. Changes apply in every mode without the typed confirmation.

In the dashboard: After the bot exits on its own, the launcher waits this long before restarting it (read at launcher start).

What it does

When the bot exits on its own (a crash, not Ctrl+C), the launcher alerts you and waits this long before restarting it. Read when the launcher starts.

Example

At 10 seconds: the bot crashes at 03:00:00, you get a "BOT EXITED" alert with the last log lines, and the bot starts again at about 03:00:10.

When to change it

Raise it if crashes come from a problem that needs time to clear (for example a network outage). A change takes effect the next time you start the launcher.

Higher, lower, on or off

Higher: slower restarts, less chance of a fast crash loop. Lower: the bot is back sooner.

Risk notes

Low: while the bot is down, exchange stops still protect open positions, but nothing is managed.

Launcher: restarts per hour

launcher_max_restarts_per_hour needs restart

Default
5
Range
1 to 100
Type
Whole number
Live confirmation
No. Changes apply in every mode without the typed confirmation.

In the dashboard: More automatic restarts than this within an hour and the launcher gives up and alerts you.

What it does

Limits automatic restarts. When this many restarts have already happened within the last hour and the bot exits again, the launcher gives up, sends an alert and stops. Read when the launcher starts.

Example

At 5: the bot crashes six times between 10:00 and 10:40. The first five crashes are each followed by a restart; at the sixth, the launcher gives up and alerts "launcher gave up". You must start it again by hand.

When to change it

Lower it if you prefer to look at crashes yourself sooner. A change takes effect the next time you start the launcher.

Higher, lower, on or off

Higher: more attempts to recover on its own, but a broken bot can keep restarting. Lower: gives up sooner and leaves it to you.

Risk notes

Low to medium: after giving up, open positions keep their exchange stops but are no longer managed.

Launcher: clean stop on Linux signals

launcher_posix_signals needs restart

Default
On
Range
On or off
Type
On/off
Live confirmation
No. Changes apply in every mode without the typed confirmation.

In the dashboard: Linux/macOS: the bot runs in its own process group; Ctrl+C, SIGTERM (kill) and SIGHUP (the terminal or tmux session closes) all reach the launcher, which stops the bot cleanly (read at launcher start).

What it does

On Linux (the VPS) the launcher starts the bot in its own process group and handles Ctrl+C, SIGTERM and SIGHUP itself: each one makes the bot stop cleanly (safe stop, clean-stop flag), exactly like Ctrl+C on Windows. Ignored on Windows. Read when the launcher starts.

Example

On: the SSH connection drops and the tmux session is killed by mistake; the launcher gets SIGHUP (signal 1), passes a Ctrl+C to the bot, logs the safe-stop lines and exits within about 30 seconds. The next start reports a clean stop.

When to change it

Leave it on. Turn it off only to get the old behaviour where the terminal's signals go to the bot directly.

Higher, lower, on or off

On: every way of stopping the bot on Linux is a clean stop. Off: closing the terminal can kill the bot without its safe stop.

Risk notes

Low. Off is riskier: an abrupt stop can leave an order action half done (reconciled at the next start).

Launcher: clean stop on Ctrl+Break (Windows)

launcher_ctrl_break needs restart

Default
On
Range
On or off
Type
On/off
Live confirmation
No. Changes apply in every mode without the typed confirmation.

In the dashboard: Windows: Ctrl+Break in the bot's window is handled like Ctrl+C: the bot stops cleanly, the shutdown lines are logged and the launcher does not restart it (read at launcher start).

What it does

On Windows, a Ctrl+Break sent to the bot's console window (used when a window ignores Ctrl+C) is handled by the launcher exactly like Ctrl+C: the bot shuts down cleanly (safe stop, clean-stop flag), its shutdown lines are written to the log, and the launcher exits without restarting it. Ignored on Linux. Read when the launcher starts.

Example

On: a 8787 window started by a script ignores Ctrl+C; the deploy sends Ctrl+Break, the log ends with 'Application shutdown complete' and '[launcher] stopped by Ctrl+C', and the next start reports a clean stop.

When to change it

Leave it on.

Higher, lower, on or off

On: Ctrl+Break is a logged clean stop. Off: Ctrl+Break ends the launcher at once; the bot still stops cleanly but its last lines are missing from the log.

Risk notes

None.

Log file: rotate at

log_rotate_mb needs restart

Default
50 MB
Range
0 to 10000 MB
Type
Number
Unit
MB
Live confirmation
No. Changes apply in every mode without the typed confirmation.

In the dashboard: When the log file reaches this size the launcher renames it to .1 (.1 to .2 ...) and starts a new one. 0 = never rotate (read at launcher start).

What it does

Size at which the launcher rotates the log file (logs/wallbot.log or logs/wallbot_live.log): the current file becomes .1, .1 becomes .2 and so on. 0 = never rotate. Read when the launcher starts.

Example

At 50 MB with 5 kept: the Trading log reaches 50 MB at 14:00; it is renamed wallbot_live.log.1 and a new wallbot_live.log starts. At most about 300 MB of log is kept.

When to change it

Raise it if you want longer history in one file; lower it on a small disk.

Higher, lower, on or off

Higher: fewer, larger files. Lower: smaller files, rotated more often.

Risk notes

Low. With 0 the log grows without limit and can fill the disk budget.

Log file: old files kept

log_keep needs restart

Default
5
Range
1 to 100
Type
Whole number
Live confirmation
No. Changes apply in every mode without the typed confirmation.

In the dashboard: How many rotated log files are kept; older ones are deleted.

What it does

How many rotated log files (.1 ... .N) are kept; the oldest is deleted at each rotation. Read when the launcher starts.

Example

At 5: wallbot_live.log.1 to wallbot_live.log.5 are kept; at the next rotation the old .5 is deleted.

When to change it

Raise it to keep more history (it counts toward the disk budget).

Higher, lower, on or off

Higher: more history, more disk. Lower: less of both.

Risk notes

Low: older log lines are gone once deleted (trades and events stay in the database).