← All results
Positive test

AI-assisted 8am Candle Reversal

The video reports a practice-challenge pass from January 2 to May 7, 2025. It also shows a live trade and talks about a longer coded test.

Source Eddy PipsVideo published Mar 27, 2026
Watch the video
Scorecard · Challenge replaySample size not stated
Trades
—
Win rate
—
Profit factor
—
Reward : risk
1R
Biggest drop
—
Net result
—

A dash means the video did not report that number.

Two ways to check this test

Open the actual trade log

Every trade from my test, in the order it happened. The numbers on this page come from these trades.

My breakdown

What this is

This is a strategy I built myself, using ChatGPT as a research assistant. AI didn't hand me a profitable strategy. It helped me organize ideas, write the rules down tightly, and test variations faster. The strategy helped me pass a prop firm challenge, and in a replay from January 2 to May 7, 2025, it passed both phases of a practice challenge.

It was tested on US500 (the S&P 500 CFD), and the same idea can be applied to MES or ES futures.

The rules

Step 1: the 8 AM candle's size. At 9:00 AM New York time, measure the 8:00 AM 1-hour candle from high to low. It must be between 10 and 24 points. Smaller means the market is too quiet. Bigger means the move is overextended. Either way, skip the day.

Step 2: where it closed. Draw a Fibonacci from the candle's high to low with levels at 0, 0.45, 0.55 and 1. The candle must close above 0.55 or below 0.45. A close in the middle is an indecision candle. Skip it.

Step 3: the first break. Mark the candle's high and low and drop to the 1-minute chart. Wait for price to break one side. Breaks the high first: shorts only. Breaks the low first: longs only. If the first candle after 9:00 breaks both, skip the day.

Step 4: close back inside. Price has to close back inside the range within 60 minutes of the break. No close back inside by then, no trade.

Step 5: the blackout window. If that close back inside happens 2 to 5 minutes after the break, skip the trade. Closes in that window weakened the edge in testing.

Step 6: don't enter yet. Closing back inside only says the breakout might be failing.

Step 7: retest rejection. Wait for price to come back to the broken level (touching or going past it) and then close back inside again. This has to happen within 2 hours.

Step 8: entry. Enter at the open of the next 1-minute candle after the rejection candle closes. 12-point stop, 12-point target (1R). Maximum hold six hours.

What the testing showed

  • The challenge replay (January 2 to May 7, 2025) passed both phases. It took months, which is the trade-off for a strategy with this many filters.
  • The live trade on March 20 that pushed my real challenge over the line is broken down step by step in the video: a 16.89-point 8 AM candle, a close above 0.55, a break of the high at 9:01, a close back inside at 9:02, and a retest rejection that triggered a short at 9:27.
  • I also discuss a longer coded test in the video. I prefer a risk of about 1.25% per trade in simulation, but treat each run separately.

A live pass isn't a funded result, and one challenge isn't proof. But the rules are precise enough that you can test them yourself.

Why all the filters?

Each one removes a type of bad day: the range filter removes quiet and wild mornings, the close-position filter removes indecision, and the blackout window removes the fast fake-outs. That's the part where AI helped most: testing which filters actually made a difference instead of guessing.

Takeaway

If you want to use AI for trading, use it like this: to make your rules precise and test them, not to invent a strategy and trust it. My guide to how I backtest a trading strategy covers the process, and the ChatGPT indicator tutorial shows another way to use it.

The test

How I tested
Replay and coded research
When
January 2–May 7, 2025 challenge replay
Market
US500 / XPS500 as described in video
Trades
More than one stretch of trades. See below.
Risk per trade
1.25% discussed as a preferred simulation setting; individual runs must be distinguished
Fees
Not said in the video

Trade log

AI-assisted 8am Candle Reversal

Download PDFDownload endpoint checked 2026-09-22. Contents may differ from the video; see source notes.

More from the test

Instrument
US500 / XPS500 as described in video
Test window
January 2–May 7, 2025 challenge replay
Risk per trade
1.25% discussed as a preferred simulation setting; individual runs must be distinguished
Stop and target
12 points / 12 points; six-hour maximum hold

A note

A live pass in the video is not a funded result.

Where the numbers come from

Checked Sep 22, 2026. These numbers come from the video. They are backtest or simulated results, not live trading, and not financial advice.

Original video ↗

Test your own trades

EdgeFinder

The app I built to log a backtest and run it through prop firm challenge rules. It is free to try.

Try EdgeFinder free ↗

Backtest Lab

Members get the code behind my backtests, the results before the video goes public, and a vote on what I test next.

Join Backtest Lab ↗

EdgeFinder is my app and Backtest Lab is my channel membership, so both of these pay me. They never change a test result.