Performance Statement
We report the way a professional desk does: our own trading record first, then the published backtest behind it, each measured one way and labelled for what it is. Every number on this page is recomputed from our own trading record and our published backtest, normally once an hour; nothing is typed in by hand.
Every figure here can be reproduced from the published record and the method described at the bottom of the page, and where a number is modeled rather than measured, the page says so.
Updated Sep 9, 2026, 1:42 AM ET; backtest published through Sep 6, 2026.
At a glance
Longest window first: the multi-year backtest that came before all of it, then the last 12 months, the last 90 days and the last 30 days of our own trading. Every rate is wins over wins plus losses. Dollar results are intentionally not shown. Position size, broker and fill quality are each member's own, so a single dollar figure would describe nobody's account.
A strategy is judged over hundreds of trades, not over a few weeks. The 30-day and 90-day tiles show recent activity, and a sample that small can swing from month to month by chance alone, before any change in the strategy; the shaded bar under each tile is the statistical range (95% confidence) for a sample that size, which is how far the measured rate can sit from the true one. The 3-year backtest and the full forward record carry the weight. What those stretches actually looked like at two weeks, three months and a year is the section directly below.
By strategy
Win rate with the number of trades that ended in a win or a loss in brackets, longest window first. Where a window holds none of those, the cell reads n/a. Hover a cell for the statistical range at that sample size. The backtest column exists only for the two strategies that publish one, ES/SPY and NQ/QQQ; AITrader builds each signal live, so it is judged on the forward record alone.
| Strategy | Backtest | 12 months | 90 days | 30 days |
|---|---|---|---|---|
| ES/SPY Options Strategy | 75.2% (512) | 82.5% (57) | 78.3% (46) | 53.8% (13) |
| AITrader | no published backtest | 64.9% (114) | 64.9% (114) | 70.5% (88) |
| NQ/QQQ Options Strategy | 83.1% (682) | 88.4% (43) | 88.4% (43) | 80.0% (10) |
What a normal stretch looks like
The same published backtest, measured over the lengths a member lives through rather than over one bad week. ES/SPY first, then NQ/QQQ.
ES/SPY, stretches that finished ahead
In our record, 40% of two-week stretches were flat or down. 99% of twelve-month stretches finished ahead.
The longest stretch without a new high in the record was 239 days (Nov 2, 2023 to Jun 28, 2024), with a drawdown of 8.0R inside that stretch.
Every calendar day in the record is a possible start date; each bar counts the stretches of that length that finished above where they started, in modeled R over the stages we trade today.
NQ/QQQ, stretches that finished ahead
Your first year
We ask every member to judge the strategy at three points, not at two weeks.
In the record, 84% of three-month stretches finished ahead, typical result +4.5R. Look at your win rate and your trade count next to ours, not at a single week.
87% of six-month stretches in the record finished ahead. At six months, compare the shape of your results with ours: many small gains against fewer full-premium losses.
99% of twelve-month stretches finished ahead. A year is the horizon this strategy is designed to be judged on.
Backtest of record
HypotheticalOne backtest, published once and shown by every AlgoIndex surface: this page, the member area and the mobile app all read the same file. ES/SPY over Jan 1, 2023 to Sep 6, 2026, 513 signals across the two stages currently traded. Stages 3 and 4 (Mid-Day) were retired and are excluded, so this describes what runs today rather than a blend with a strategy nobody trades. Backtest outcomes are modeled from the exit rule, not from broker fills: any exit with the market in the trade's favor, at the target or at the close, counts as a gain of half the premium risked; any exit against the trade counts as the premium lost; a flat exit counts as zero. Results are shown in R, where one R is the premium risked on a single trade. Drawdown is measured on the end-of-day cumulative result, so a swing that opens and closes inside a single day is not counted in it.
Cumulative return (R)
Monthly return (R)
| Jan | Feb | Mar | Apr | May | Jun | Jul | Aug | Sep | Oct | Nov | Dec | Year | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 2023 | +2.5 | +2 | +4.5 | +1.5 | +3 | 0 | -1.5 | +3 | -0.5 | +3.5 | -2.5 | +0.5 | +16R |
| 2024 | -2 | -2.5 | +1.5 | +3 | 0 | +3 | +3 | +5.5 | +3 | 0 | +3.5 | -1 | +17R |
| 2025 | +2 | -1 | +2 | +4 | +3 | +2 | +4 | +3 | -2.5 | +1 | 0 | +2 | +20R |
| 2026 | -5.5 | +4 | +7.5 | +1.5 | +1 | +7.5 | 0 | -2 | -1 | +13R |
Trade outcome distribution
The model has three outcomes and no others: an exit with the market in the trade's favor, whether at the target or at the close, an exit against the trade, or a flat exit. Many small gains against fewer full-premium losses is the shape the strategy is built to produce.
By stage
Both stages we trade today run in the New York morning session: a long and a short.
| Stage | Signals | Wins | Losses | Win rate |
|---|---|---|---|---|
| S1 NY AM Long | 311 | 234 | 76 | 75.5% |
| S2 NY AM Short | 202 | 151 | 51 | 74.8% |
| Current stages combined | 513 | 385 | 127 | 75.2% |
By year, current stages
| Year | Signals | Wins | Losses | Win rate |
|---|---|---|---|---|
| 2023 | 111 | 84 | 26 | 76.4% |
| 2024 | 127 | 96 | 31 | 75.6% |
| 2025 | 141 | 107 | 34 | 75.9% |
| 2026 | 134 | 98 | 36 | 73.1% |
Every AITrader signal is built live from real-time options-flow and market data, so there is no fixed historical parameter set to simulate. It is judged purely on its forward record below.
NQ/QQQ runs the ES/SPY logic scaled to the Nasdaq-100. Its published backtest covers 683 signals at a 83.1% win rate under the same payoff model, and its forward record is still building.
Forward record since Jun 1, 2026, all three strategies
Recorded forward247 trades taken on our own account and recorded as they happened, one row per trade. Since Aug 14, 2026 these trades run on a live brokerage account; before that the same signals were forward-tested with simulated fills. Win rate counts only trades that ended in a win or a loss. Trades that closed flat, whose order did not fill, or whose result is not yet recorded are counted in the totals and left out of the rate. Cancelled signals that never became a trade are not counted at all. Dollar results are intentionally not shown. Position size, broker and fill quality are each member's own, so a single dollar figure would describe nobody's account.
Updated Sep 9, 2026, 1:42 AM ET; backtest published through Sep 6, 2026.
(47 wins / 10 losses)
Live account since Aug 14: 11 trades, 45.5% (5 wins / 6 losses). 9 trades closed flat, did not fill, or has no result recorded yet: counted in the total, left out of the rate.
(74 wins / 40 losses)
Live account since Aug 14: 38 trades, 70.3% (26 wins / 11 losses). 23 trades closed flat, did not fill, or has no result recorded yet: counted in the total, left out of the rate.
(38 wins / 5 losses)
Live account since Aug 14: 9 trades, 77.8% (7 wins / 2 losses). 1 trade closed flat, did not fill, or has no result recorded yet: counted in the total, left out of the rate.
Month by month
Every month of the forward record, per strategy, dated by the day the trade closed.
| Month | ES/SPY Options Strategy | AITrader | NQ/QQQ Options Strategy | All strategies |
|---|---|---|---|---|
| Jun 2026 | 100% (28) | 0% (6) | 78% (9) | 91% (43) |
| Jul 2026 | 83% (18) | 48% (40) | 95% (20) | 73% (78) |
| Aug 2026 | 54% (16) | 69% (83) | 91% (11) | 69% (110) |
| Sep 2026 | 75% (4) | 88% (8) | 75% (4) | 81% (16) |
Win rate for the month with the number of trades in brackets. A month of a dozen trades is a thin sample: read the row, not the best cell in it.
Our strategies
Three strategies, each reported the same way: a multi-year backtest where the logic is rules-based and reproducible, and a forward record on our own account for all three.
(66 trades so far)
- Signal
- ES (S&P 500) futures
- Vehicle
- SPY options
- Structure
- Single-leg long, 0DTE
- Stages
- Stage 1 + Stage 2 (NY AM)
The flagship automated strategy. ES futures trigger the signal and SPY options are the execution vehicle, in two New York morning stages.
(137 trades so far)
- Signal
- Live multi-source analysis
- Vehicle
- SPY options
- Structure
- Single-leg long (call or put), 0 to 1DTE
- Stages
- On demand
AI-driven signals built at the moment from real-time options-flow and market data. Because each signal is built live rather than from a fixed parameter set, AITrader is judged on its forward record.
(44 trades so far)
- Signal
- NQ (Nasdaq-100) futures
- Vehicle
- QQQ options
- Structure
- Single-leg long, 0DTE
- Stages
- Stage 1 + Stage 2 (NY AM)
The ES/SPY logic generalized to the Nasdaq-100, with its own multi-year backtest and a forward record that is still young.
How we count
One trade, counted once
One trade is a single round trip: an entry signal, a position held, and an exit at the profit target, at the end of the session, or at expiry, whichever comes first. If an outcome is corrected later, the correction and the trade it corrects are the same trade and are counted once, so a trade can never appear twice in a total.
How the win rate is calculated
Win rate is wins divided by wins plus losses. Trades that closed flat, whose order did not fill, or whose result is not yet recorded are counted in the totals and left out of the rate. Cancelled signals that never became a trade are not counted at all.
No dollar figures
Dollar results are intentionally not shown. Position size, broker and fill quality are each member's own, so a single dollar figure would describe nobody's account. What we can show plainly is the count of trades, the outcome of each one, and the rate they produce.
One published backtest, and an hourly page
The backtest runs the production strategy logic over historical 5-minute futures bars, limited to the stages we trade today. It is a simulation, not live trading, and it does not model every real-world cost or fill. Backtest outcomes are modeled from the exit rule, not from broker fills: any exit with the market in the trade's favor, at the target or at the close, counts as a gain of half the premium risked; any exit against the trade counts as the premium lost; a flat exit counts as zero. It is one published record shared by the website, the member area and the app, and this page recomputes it, and our own record, at most once an hour.
Updated Sep 9, 2026, 1:42 AM ET; backtest published through Sep 6, 2026. Operational partners (execution and charting) are listed in the site footer.
Important note on performance
Past performance is not a guarantee or reliable indicator of future results. The backtest figures on this page are hypothetical and were generated with the benefit of hindsight; hypothetical results have inherent limitations and do not represent actual trading. The forward figures cover a short, early sample from a single account and may not be representative over time. Trading options involves substantial risk, including the risk of losing the full amount risked. Nothing here is investment advice or a solicitation. For the full disclosure, see our Performance Disclaimer and Risk Disclaimer.