Last updated: October 2026
Can an automated futures trading strategy actually progress from simulation testing to a prop-firm evaluation, a funded trading account and eventually real payouts?
That's what I'm trying to find out.
I've spent a lot of time trading futures, testing different strategies and developing automated trading systems. Along the way, I've become increasingly interested in a simple idea: instead of trying to manually capture every small movement in the market, could I build a strategy that takes trades automatically, controls its risk and gives itself enough room to capture the occasional large move?
This page documents that experiment.
I'm starting with the Micro E-mini Nasdaq-100 futures contract (MNQ) and an automated strategy I've developed for NinjaTrader.
Rather than showing a backtest after I've already optimized everything and presenting the finished result, I want to document what happens as I actually test and develop the strategy.
That includes the good results.
It also includes losing periods, drawdowns, strategy changes, failed ideas and potentially failed prop-firm evaluations.
The initial simulated results have been encouraging, but this experiment is still at a very early stage.
My first published five-day test produced $2,019.50 in simulated net profit from 46 trades, using two MNQ contracts.
That does not mean I've discovered a money-making machine.
It means I've found something interesting enough to continue testing.
This page will be updated as the experiment develops.
Current Automated Trading Experiment (Updated October 2026)
Here is where the experiment currently stands.
|
|---|
Whenever I make a material change, I'll document it on this page rather than quietly replacing the old results.
NEW: October 2026 Update — Why I Stopped Trading 23 Hours a Day?
My original experiment allowed the automated futures trading strategy to operate for approximately 23 hours per day.
The reasoning seemed sound at the time.
Rather than backtesting dozens of different sessions and selecting whichever hours happened to produce the best historical results, I wanted to expose the strategy to almost the entire futures trading day and collect data.
The initial SIM results were excellent.
From September 14–18, the strategy produced +$2,019.50 in simulated profit from 46 trades using two MNQ contracts, with a maximum drawdown of approximately $301.
But once I moved beyond that initial testing period, I discovered a major weakness in the 23-hour approach.
The 23-Hour Trading Window Failed the Evaluation Test
The strategy was moved onto an evaluation account, and the result was very different from the initial SIM week.
On the first day, the strategy experienced a series of losing trades and ultimately reached the account's full daily loss limit.
That was the first major warning.
The following day, I allowed the experiment to continue. Before midnight, the strategy lost the remaining available account balance and the evaluation was effectively over.
The account lasted two days.
That's obviously very different from the five consecutive profitable days produced during the original SIM test.
But it's also exactly why I wanted to document this experiment publicly.
A good week of simulated results doesn't prove that a strategy is robust.
Moving the same idea into a different period and allowing it to operate under actual evaluation constraints exposed a weakness very quickly.
What Went Wrong?
Looking through the trades, one problem became increasingly obvious: 23 hours of market exposure was simply too much for this version of the strategy.
The strategy wasn't only participating when Nasdaq produced clean directional movements.
It was also continuously available to trade during quieter, choppier and less productive periods.
That meant it could respond to what initially appeared to be a move, enter a position and then get caught when that movement quickly failed.
One false move isn't particularly important.
The problem is repeatedly exposing the strategy to them throughout almost the entire trading day.
My original assumption was that allowing the strategy to trade for longer would give it more opportunities to catch the large moves I'm looking for.
What I hadn't adequately accounted for was the other side of that equation:
More opportunity also means more exposure to bad opportunities.
A strategy doesn't necessarily become better because I give it 23 hours to find trades.
Why I Didn't Abandon the Strategy
After the evaluation failed, I had two choices.
I could conclude that the automated strategy didn't work.
Or I could look at why it failed and test whether there was a logical adjustment worth investigating.
I chose the second option.
Importantly, I didn't want to completely redesign the strategy simply because of two bad days.
The entry philosophy, wider initial stop and tick-based trade management were still ideas I wanted to investigate.
Instead, I focused on one major variable:
Time.
From Tuesday around midnight, I returned to backtesting with a much more finite trading window.
Instead of allowing the strategy to operate approximately 23 hours per day, I restricted it to:
12:00 am to 6:00 am.
That's only a six-hour session.
Why Midnight to 6:00 AM?
The purpose of this change isn't to claim that midnight to 6:00 am is magically the best time to trade Nasdaq.
I don't have enough evidence to make that conclusion.
The objective is much simpler.
I want to reduce the amount of time the strategy is exposed to the market and determine whether a more controlled trading window can retain enough of the profitable moves while eliminating some of the unnecessary trades that damaged the 23-hour version.
This is an important distinction.
I'm not trying to find the six historically perfect hours and then present an optimized backtest.
I'm testing whether less market exposure produces a more robust strategy.
That means I'm now interested in more than net profit.
I want to compare:
- Total number of trades
- Maximum drawdown
- Consecutive losses
- Average winner and loser
- Profit factor
- Number of full-stop losses
- Large winning trades captured
- Profit generated per trade
- Profit generated relative to drawdown
If the six-hour version makes less money but experiences dramatically lower drawdown, that could actually be an improvement.
The First Major Lesson From the Experiment
This evaluation failure has already taught me something useful:
More trades aren't necessarily better.
My original 23-hour test looked fantastic on paper.
It produced more than $2,000 in simulated profit in five days and gave the strategy almost the entire futures trading day to find opportunities.
But the evaluation demonstrated how quickly that constant exposure could become a liability when market conditions changed.
So the strategy has now entered its next development stage:
Version 1: Approximately 23 hours per day
Result: Strong initial SIM performance, but evaluation failed on Day 2.
Version 2: Midnight to 6:00 am only
Objective: Reduce low-quality market exposure while preserving the strategy's ability to captu