One condition. No indicator. No parameter to optimize. It filtered 80% of bad trades from a NASDAQ breakout strategy.
The condition: is the high of the entry bar above today's open?
That's it. One line of code.
If yes, the market has been trading above today's open. There's an established uptrend for the day. Enter the long breakout.
If no, the market is below today's open when your signal fires. You're not in an uptrend. You're trading noise. Skip.
Before:
- Net profit/drawdown ratio: 5.24
- Drawdown: ~$40,000
- Average trade: $120
After:
- Net profit/drawdown ratio: 7.97
- Drawdown: ~$25,000
- Average trade: $162
Net profit went up while drawdown got cut. That's the signature of a genuine filter. It's removing the right trades, not random ones.
Why does it work? Breakout trading is about momentum. If you're buying a breakout to the upside but the market has been trading below its open all day, there's a contradiction. The price level is breaking out, but the day's direction says otherwise. Those trades are the ones that chop you up.
No parameters means no overfitting. This is pure trading logic. You could test it on any long breakout strategy, on any market, in minutes.
The power isn't in complexity. It's in asking the right question.
What's the simplest filter you've ever used?
The Strategy Starter Playbook provides the exact formula behind nearly 2,500 hedge-fund strategies, plus 11 example strategies built from it across markets such as gold, NASDAQ, Dow, S&P, and Nikkei. Each example includes the full logic. See the process here:
https://t.co/4DkjNGGgD9
The gold edge was not in the main session.
It showed up after 4:05pm Chicago
on Monday and Thursday only.
Gold futures. 25-minute bars. Long only.
Exact rules:
> Point of initiation: 20-period SMA
> Breakout: SMA + 2.5 × ATR(14)
> Window: 4:05pm to 11:50pm
> Exit: 11:50pm
On that prototype:
> Win rate: 56%
> Average trade: $261 per contract
> About 2 to 2.5 trades a month
2.25 and 2.75 also held up.
So 2.5 is not a single-point fit.
Main session and overnight had no usable long bias.
This is a foundation, not a finished strategy.
I ran 16.5 million filter tests on NASDAQ.
One condition beat every other oscillator, trend indicator, volatility filter, and volume condition.
The winner: a simple CCI correction filter.
Exact rule for long entries:
Lowest(CCI(27), 18 bars) < –100
In plain English:
Only take the long breakout if, somewhere in the last 18 bars, CCI(27) dipped below –100.
(i.e. the market has already had a real correction)
Results across 1,000 strategy variations:
> Net profit improved on 69% of strategies
> Average trade improved by 22% on more than half
> Bounce index: 57.8% (turned that many losing variations profitable)
On the base NASDAQ breakout:
> Profit/drawdown ratio: 7.53 → 8.6
> Average trade: $116 → $149
Breakouts taken after a genuine washout have room to run.
Breakouts taken into already-extended strength tend to fade.
This wasn’t tested on one strategy.
It was stress-tested across 1,000 different ones.
Most traders guess the breakout level.
There is a better way.
A breakout entry only needs two measurable parts:
1. Point of initiation
- The reference level the calculation starts from.
- It can be a moving average, a pivot, or the high/low of a lookback window.
2. Space
- How far price must travel beyond that level.
- Measured with ATR so it adapts to volatility.
Long entry = Point of initiation + (ATR × multiplier)
Short entry = Point of initiation – (ATR × multiplier)
The multiplier controls how selective the entry is.
Lower multipliers create more trades. Higher multipliers create fewer but cleaner ones.
One tested candidate on E-mini NASDAQ used a 6 a.m. pivot reset, ATR period 5, and a 1.0 multiplier.
It scored 68% on robustness.
No drawing on charts.
No subjective levels.
Just a defined reference plus a volatility-adjusted distance.
Want to know why your strategy looks amazing in backtests… but fails in real trading?
You’re probably using too many or too few filters.
We tested this on over 2,500 trading strategies.
Here’s what the data revealed:
1 optimizable input = weak true OOS performance
2-3 optimizable inputs = strong, stable real performance
4 optimizable input = weak true OOS performance
5-6 inputs = strong, stable real performance
More than 6 inputs = huge drawdowns and unstable equity curves
Why does this happen?
To be honest… we don’t know exactly.
But this is what the data tells us and we always follow the data.
So:
✅ 2–3 or 5–6 inputs = more robust
✅ 1 optimizable input doesn’t make the strategy robust enough
✅ More than 6 inputs = overfitting and instability
Not sure about 4 inputs, could be just an anomaly.
Most traders guess the breakout level.
There is a better way.
A breakout entry only needs two measurable parts:
1. Point of initiation
- The reference level the calculation starts from.
- It can be a moving average, a pivot, or the high/low of a lookback window.
2. Space
- How far price must travel beyond that level.
- Measured with ATR so it adapts to volatility.
Long entry = Point of initiation + (ATR × multiplier)
Short entry = Point of initiation – (ATR × multiplier)
The multiplier controls how selective the entry is.
Lower multipliers create more trades. Higher multipliers create fewer but cleaner ones.
One tested candidate on E-mini NASDAQ used a 6 a.m. pivot reset, ATR period 5, and a 1.0 multiplier.
It scored 68% on robustness.
No drawing on charts.
No subjective levels.
Just a defined reference plus a volatility-adjusted distance.
Most traders only look at the midnight pivot.
I tested nearly 300 pivot point strategies on E-mini NASDAQ.
The midday version ranked higher on robustness.
Not the conventional boundary.
The one that sits in the middle of the session.
Best candidate used:
> Midday pivot as the point of initiation
> ATR period 20
> Multiplier around 1.4
Nearly 90% of the surrounding multipliers stayed profitable.
It held up on unseen data and ranked at the top in walk-forward.
The edge survived small shifts in the multiplier.
It was not a single optimized peak.
Most traders stick with the standard midnight pivot because it is familiar.
This test showed the midday boundary carried more consistent results across the variations.
Once you have the right point of initiation, you can measure the space with ATR and keep the rest of the model simple.
I’ve used every indicator you can think of.
But few are as practical, consistent, and scalable as ATR (Average True Range).
It’s not just for measuring volatility.
It can power your entire strategy.
Here’s how I use it across system development:
Entries
➡️ Start with a POI (Point of Initiation).
➡️ Multiply ATR by 2.5 to set your breakout entry.
➡️ It adapts to market conditions and removes the guesswork.
Filters
➡️ Compare two ATR periods to spot low-volatility zones.
➡️ It keeps you out of weak setups and sharpens your edge.
Exits
➡️ Let ATR guide your stop-loss placement.
➡️ It responds to real volatility, not emotion.
ONE indicator.
THREE essential roles.
THOUSANDS of ways to customize and scale.
If you want structure without complexity, THIS is where to BEGIN.
Watch the video and grab the free resource below ⬇️
We built a slope filter with AI.
It crushed on NASDAQ.
But the real test?
Cross-market performance.
So we ran it against:
ES — big jump in performance
RTY & EMD — small but positive changes
YM & Nikkei — worse
No TUNING.
No OVERFIT logic.
4 out of 6 markets saw gains.
This is how you test AI-generated strategies with LOGIC, STRUCTURE, and LIMITS.
Is it magic? NO.
Is it usable? DEFINITELY.
Full walkthrough + free strategy in the comments.
Most traders only look at the midnight pivot.
I tested nearly 300 pivot point strategies on E-mini NASDAQ.
The midday version ranked higher on robustness.
Not the conventional boundary.
The one that sits in the middle of the session.
Best candidate used:
> Midday pivot as the point of initiation
> ATR period 20
> Multiplier around 1.4
Nearly 90% of the surrounding multipliers stayed profitable.
It held up on unseen data and ranked at the top in walk-forward.
The edge survived small shifts in the multiplier.
It was not a single optimized peak.
Most traders stick with the standard midnight pivot because it is familiar.
This test showed the midday boundary carried more consistent results across the variations.
Once you have the right point of initiation, you can measure the space with ATR and keep the rest of the model simple.
@guyonabuffalo79 You're absolutely right! Overfitting is a serious risk, especially with AI. That's why I tested the idea across multiple markets and out-of-sample data. AI is great for generating ideas, but it doesn't replace proper validation.
This AI trading idea made me look like a fool.
I've been calibrating the ADX indicator for over 20 years. Thousands of parameter combinations across every major index market. So when I asked AI to improve my ADX filter on an E-mini NASDAQ strategy, I expected something marginal.
It proposed replacing part of my logic entirely.
Original strategy: $87K net profit, $23K drawdown.
After AI's filter: $200K+ net profit, $13K drawdown.
Drawdown cut nearly in half. Profit more than doubled.
The strangest part? The AI's suggestion was simpler than what I had. Not more complex. Simpler. It took one piece of the ADX logic and replaced it with a cleaner, more direct measurement of what ADX was trying to capture in the first place.
This doesn't mean AI replaces experience. I still had to evaluate whether the improvement was genuine or just overfitting. I still ran it across multiple markets and out-of-sample data.
But AI spotted a pattern in indicator behavior that 25 years of screen time missed. When that happens, you'd be foolish to ignore it.
What's your experience using AI to improve strategies?
@Einherjar123 Thanks, Svenn! Great question. I'd suggest exploring ATR-based trailing stops and testing different exit rules to see what works best for your strategy. I've covered some useful ideas here: https://t.co/nycrcgOpFl
A strategy can look great in a backtest and still fail every prop firm check.
Here are the four tests that expose it:
1. Market vs Strategy
- Independent edge score: 76/100 on one gold model.
2. Edge Test
- Compared against 1,000 similar variants: 67/100.
3 Robustness
- Shift entry/exit by 1–3 bars: 80/100.
4. Outliers
- Remove the top 5% of trades.
- Those trades made up 137% of the total profit.
- What’s left after you remove them: negative.
Most strategies die on check 4.
You can have a rising equity curve, a solid robustness score, and still be one or two lucky trades away from a failed challenge.
Which of these four checks do you actually run before you pay a prop firm fee?
@JordanVossAI Absolutely! It's easy to make a strategy look great on past data. The real challenge is proving it can hold up beyond the data it was built on. That's where proper validation makes all the difference.
@JordanVossAI A great backtest is just the starting point. The real confidence comes from challenging the strategy, testing its weaknesses, and understanding whether the edge is truly robust.
@dolvero Exactly! That's why I consider the outlier test so important. A profitable backtest means very little if almost all the profits depend on a handful of exceptional trades. The real question is whether those opportunities are still present in the market.
You’ve heard about all kinds of exit rules.
ATR-based, trailing stops, volatility brackets...
But our hedge fund tests proved one rule beats them all:
A fixed USD profit target.
✅ It’s simple
✅ It’s consistent
✅ It’s nearly impossible to overfit
For swing strategies, we pair it with end-of-week exits.
For day trading, we add end-of-day timing.
That combo alone powers dozens of our best strategies.
And we keep seeing it win across all futures markets.
Want reliable performance?
START with reliable exits.