Kalman Filter trading strategies: backtest vs live results
A Kalman filter is an adaptive smoothing of price that reacts faster than a fixed moving average when the trend changes.
10 published strategies, 3 of them built by Trigr. Each one has its own page, where its backtest and its live paper record are shown separately when it has them. Of the 10 with 14 or more days of live paper record, 4 are up (median −0.1%, average +1.3%, over a median 70 days). In backtest, 10 of 10 were up (median +103.8%, average +213.3%). The average adds every return, so a few very large ones move it. A backtest covers years and a live record covers weeks, so the two returns are not the same length of time.
The median is the middle return. The average adds every return, so a few very large ones move it.
Backtest against live, strategy by strategy
One dot per strategy with 14 or more live days: how good its backtest looked, against what it has done since.
Live returns
How the 10 live paper returns are spread. Each record runs from the day its strategy was published, so they cover different lengths of time.
Backtest and live results by indicator
Each row counts the strategies of one group: how many are up in backtest, and how many of those with a live record are up, flat or down since. The median is the middle return; the average adds every return, so a few very large ones move it.
| Indicator | Up in backtest | ||||||
|---|---|---|---|---|---|---|---|
| Rate of Change | 2 | 2 of 2 | +19.4% | +19.4% | 0 of 2 | −0.1% | −0.1% |
| RSI | 2 | 2 of 2 | +19.4% | +19.4% | 0 of 2 | −0.1% | −0.1% |
Does a stronger backtest hold up live?
Grouped by the Sharpe ratio of the backtest: does the better-looking group do better live?
| Backtest Sharpe | Strategies | Up in backtest | Up live | Median live | Avg live | ||
|---|---|---|---|---|---|---|---|
| under 0.5 | 4 | 4 of 4 | +32.7% | +38.3% | 2 of 4 | 0.0% | +0.1% |
| 0.5 to 1 | 3 | 3 of 3 | +138.5% | +265.2% | 0 of 3 | −2.9% | −4.8% |
| 1 to 1.5 | 3 | 3 of 3 | +383.6% | +394.8% | 2 of 3 | +11.0% | +9.1% |
| 1.5 to 2 | 0 | n/a | n/a | n/a | none yet | n/a | n/a |
| 2 and over | 0 | n/a | n/a | n/a | none yet | n/a | n/a |
All 10 Kalman Filter strategies
Longest live record first, not best return first. Select a column to sort by it; open a strategy for its full record.
| Indicators | |||||||
|---|---|---|---|---|---|---|---|
| Micron Smoothed Trend Rider · Hourly MU · 1H · Long/short | Kalman Filter | +18.8% | 0.26 | −21.6% | 1947 | −9.9% | 106 |
| Gold Small-Size Trend Rider · Daily GOLD · Daily · Long only | Kalman Filter, RSI, Rate of Change | +11.6% | 0.52 | −3.4% | 147 | −0.1% | 105 |
| Gold Trend Rider · Daily GOLD · Daily · Long only | Kalman Filter, RSI, Rate of Change | +27.1% | 0.46 | −7.6% | 155 | −0.2% | 105 |
| BTC Adaptive Regime 1D v1 — 100% at 1x BTC · Daily · Long/short | Kalman Filter | +645.5% | 0.82 | −48.8% | 270 | −2.9% | 70 |
| JUP Adaptive Regime 4H v1 — 20% at 3x JUP · 4H · Long/short | Kalman Filter | +138.5% | 0.83 | −47.2% | 705 | −11.6% | 70 |
| NVDA Adaptive Regime 1D v2 — 20% at 2x NVDA · Daily · Long/short | Kalman Filter | +69.0% | 0.26 | −45.6% | 468 | +10.4% | 70 |
| STX Adaptive Regime 1D v1 — 20% at 3x STX · Daily · Long/short | Kalman Filter | +383.6% | 1.05 | −43.9% | 168 | +11.0% | 70 |
| SUI Adaptive Regime 1D v1 — 75% at 1x SUI · Daily · Long/short | Kalman Filter | +533.0% | 1.13 | −42.7% | 140 | +23.9% | 70 |
| TRUMP Adaptive Regime 4H v1 — 75% at 1x TRUMP · 4H · Long/short | Kalman Filter | +267.8% | 1.38 | −36.4% | 425 | −7.7% | 70 |
| TSLA Adaptive Regime 1D v1 — 10% at 1x TSLA · Daily · Long/short | Kalman Filter | +38.2% | 0.35 | −13.4% | 546 | +0.2% | 70 |
Questions
- How many Kalman Filter trading strategies does Trigr track?
- 10 published strategies, 3 of them built by Trigr. Each one has its own page, where its backtest and its live paper record are shown separately when it has them.
- Do Kalman Filter trading strategies hold up live compared with their backtests?
- Of the 10 with 14 or more days of live paper record, 4 are up (median −0.1%, average +1.3%, over a median 70 days). In backtest, 10 of 10 were up (median +103.8%, average +213.3%). The average adds every return, so a few very large ones move it. A backtest covers years and a live record covers weeks, so the two returns are not the same length of time.
- What is Kalman Filter in a trading strategy?
- A Kalman filter is an adaptive smoothing of price that reacts faster than a fixed moving average when the trend changes.
Keep exploring
Other topics
Figures as of 2026-11-16, refreshed hourly. Backtests run on point-in-time market data with trading and builder fees included; the live paper record is simulated on live prices from the day a strategy is published and is not a funded account. A backtest covers years and a live record covers weeks, so the two returns are not the same length of time. See how backtests are run. Past performance, backtested or paper, does not guarantee future results. Nothing here is investment advice.