Past results
Where the figures next to ready setups come from: each setup's result over its last days, worked out once a day on Prow's own server, after fees.
On this page
Prow has no backtest of your own settings. The figures you see next to a bot, such as Last 14 days: +1.0%, belong to the ready setups: once a day Prow works out what each of them made over its last days, on its own server, after fees. Continue checks your settings before the start review opens; it does not run them on past prices.
Where you see them
- Under the run length in the builder: the picked setting's result on the days it was tuned on, for example "Last 14 days: +0.6%, drawdown −0.3%, 35 trades". The settings were chosen on these same days, so this looks better than reality. Read it as a check that nothing breaks, not as what to expect. For 25+ days the line leads with the typical result of starting the same setting on other days.
- In the lists of ready setups (the Bots page, the Start a bot window and the builder's Ready setups tab): the figure on the right of each row, for example "+1.8% per 7 days". It is the typical result on later periods the setting wasn't tuned on, a fairer number. The lists are sorted by these figures, Tested setups first, and the ones your balance can't run come after. See Ready setups.
- In the start review, when you picked a run length or a ready setup: a row such as "Last 14 days as a ready setup, after fees".
The Funding bot shows its mode's result over about the last 180 days, worked out again every 6 hours. See Funding bot.
How the figures are worked out
Once a day, on Prow's own server, each ready setup (a bot type, a direction and a run length on one coin) is run over the venue's own price candles with the same code that runs your bot on Prow's servers, so orders, rounding to the venue's ticks and the minimum order size match. The days are the ones the setting was fitted on:
| Run length | Days | Candles |
|---|---|---|
| 1 to 3 days | the last 3 | 5 minutes |
| 3 to 15 days | the last 14 | 15 minutes |
| 25+ days | the last 60 | 1 hour |
- Fills: inside each candle the price is walked from the open through the low and the high to the close. A resting order fills only when the price trades through it. The order a fill triggers goes live about 30 seconds later, like the live bot's reaction time, so it misses a price that came and went in that time. A resting order takes at most a share of the volume traded near its price.
- Fees: the venue's fees (the base tier), plus the Prow fee at its standard rate of 0.025%, even during the promo. On Hyperliquid spot the Prow fee is on sells only.
- Funding (futures): paid and received at each venue's funding times.
- Slippage: market and stop orders fill worse than their price, more so in fast candles.
- Liquidation (futures): with the venue's margin rules, on isolated margin.
The three numbers behind a row are its result in % of the investment, its drawdown (the largest fall of the bot's value from a high point) and its number of trades: round trips for a grid, closed deals for DCA and Ladder. A setting is Tested when it also held up on later periods it wasn't tuned on. How it did there is the fairer figure.
What these figures leave out
- Your exact place in the order book. A real order can wait in the queue and miss a fill the figure gave it.
- Moves inside a candle. The path is guessed from the candle's open, high, low and close. When grid steps are smaller than one candle's range, several levels cross in one candle and the order of fills is a guess.
- Gaps and outages. A stop can fill much worse in a crash, or not at all while a venue is down.
- Changes in fees, funding or market rules after the period.
- Your amount. A figure is for a reference amount at the setup's own leverage, not for your investment. The same setting started on another day gives another result.
- The future. Prices that ranged last month can trend next month.
The arbitrage bot has no figure per setting; its builder shows today's spreads. Prow tested the bot on past prices: 1-hour candles and settled funding of every leg, with the costs of both legs, the liquidation guard and the cost of limit orders filling just before the price moves. See Arbitrage bot.
While bots are in beta
During the beta, the figures may be shown only to the Prow team. A ready setup then still fills in the builder and starts the same way; it just has no figure beside it.
If a check fails
Continue checks your settings against the market and the venue's limits, and says what to fix: for example the minimum investment, a stop the price is already beyond, or a market where bots aren't open. If the check itself fails, Continue stays on so you can press it again. A bot starts only from settings that have been checked.
Last updated