How results work
Every strategy is tested the same way: each trading session runs independently with $100,000 starting capital, second by second, through the same engine that powers trading accounts, settling at the 4:00 PM ET close. Because the conditions never vary, any two strategies’ results are directly comparable — the strategies list doubles as a leaderboard.
Runs are computed fee-free; your fee schedule (or the default) is applied when results are displayed. Set a custom fee schedule and you see every figure net of your own costs — applied instantly, nothing recomputes.
Results cover all trading sessions — the engine computes the full history, and that’s exactly what you see.
Results stream in session by session while a strategy runs — in the builder preview and on the strategy page alike — and appear instantly when you come back to a configuration you’ve already tried.
Orders are executed the way the platform executes them live. The price a strategy asks for is the trigger, not the price its order ends up carrying: every order goes in at the same spread-aware starting price a bot gives it, is re-priced toward a fill on the same cadence — entries rebuilt from the current market, exits chasing until they close — and fills under the same rule — three consecutive seconds for a price the market has to come to, no wait at all for one that reaches past it. What a session reports is therefore the trade the strategy would actually have gotten, chasing included. Results always use the platform’s default re-pricing cadences; tuning them per bot is a live-trading refinement.
Reading the results
A strategy’s results page reports the full metric set — total return, CAGR, Sharpe, Sortino, Calmar, max drawdown, profit factor, expectancy, win rate, and total fees under your schedule:
Every strategy reports the same set, so any two are directly comparable. Figures shown are illustrative.
Alongside the numbers come the charts to read them by: an equity curve with an SPX buy-and-hold benchmark, a drawdown view, a per-day P&L histogram, and a calendar heatmap coloring each session by its P&L.
Drilling into a session
Click any session to open its detail: that day’s orders and trades (with your fees applied), the strategy’s decision log — the running commentary of what it chose to do and why — and the second-by-second intraday chart of SPX with your order markers and the account’s equity.
The drill-in replays the session through the engine on demand, so it takes a few seconds to load; nothing about it is precomputed or stale.
Staying current
As new trading days become available, a strategy’s results fall behind — "N new sessions available". One click appends just the missing sessions; days are independent, so nothing already computed is re-run.
Reproducibility
The configuration you save is the configuration that produced the results, forever. To iterate, clone — the builder opens pre-filled, and your changes become a new strategy with its own results.
Day results are deterministic — re-running any session reproduces the stored result exactly, so the drill-in always agrees with the summary.
How to read a backtest honestly
A few habits keep full-history results from fooling you:
The equity curve’s deepest drawdown and longest flat period are what running the strategy would have felt like. If you wouldn’t have sat through it, the total return is irrelevant.
Open the P&L histogram and the calendar. One monster winning day propping up an otherwise flat strategy is a very different proposition from a steady drip of small edges — and the drill-in shows you exactly what happened on any day you don’t believe.
Everything you see is already net of your fee schedule and slippage setting. If a strategy only works with fees at zero, it doesn’t work.
The fixed $100k-per-session convention makes strategies comparable with each other; it doesn’t promise the same percentages on a different account size, where fixed per-contract fees weigh differently.