Disclaimer (read this first): Everything below is for educational purposes only. It is not investment advice, a solicitation, or a recommendation to buy or sell any futures, options, or other instruments. The performance figures discussed are simulated / hypothetical backtests from an automated research pipeline. Hypothetical results have inherent limits: they are prepared with hindsight, may not fully reflect slippage, fills, or live market impact, and past or simulated performance is not indicative of future results. Futures trading involves substantial risk of loss. Do your own work and talk to a qualified advisor before risking capital.
This morning my research stack did what it is built to do.
It pulled the overnight and early-session news flow, compressed that into an institutional-style Futures Trading Strategy Report (generated 2026-11-09 09:15:15), then spun that narrative into a small set of micro futures bots, ran them through risk rules, and produced companion portfolio and profitable-bot sheets later in the same folder (2026-11-09_091522).
People always ask me the same question when they see a PDF like this:
“Bryan, which ones are actually the most lucrative?”
So I did not rank by the pretty “expected return” ranges in the planning sheet. I ranked by the backtested net P&L the pipeline itself marked as profitable.
Spoiler: only two of six bots cleared that bar.
That is the story worth writing about — not a fantasy where every idea prints money.
What the report actually is
The PDF I started from is titled Futures Trading Strategy Report — Futures Strategies / Portfolio Metrics / Quantitative Analysis. It is explicitly stamped educational only, and the body is a news-driven institutional positioning brief across:
crypto futures and options-on-futures
rates / Treasuries / SOFR
energy and precious metals
FX and equity index futures
correlation hedges and VIX-regime rules
It is not a page of neat P&L leaderboards. It is the thesis layer: what the tape and the headlines are pushing institutions toward — Fed path, geopolitical risk premium in oil, gold as a hedge, yield-curve expression, crypto regulatory overhang, and so on.
The money question gets answered one stage later, when those themes become concrete bot specs and get backtested.
How today’s bots were born from that flow
Same run, six micro-futures scripts:
Micro E-Mini S&P 500 Fed Pivot Momentum (
MES)Micro WTI Crude Oil Geopolitical Hedge (
MCL)Micro Gold Safe-Haven Momentum (
MGC)Micro 10-Year Treasury Yield Curve Steepener (
ZN)Micro Bitcoin Regulatory Hedge (
MBT)Micro Natural Gas Seasonal Spread (
NG)
Portfolio sheet snapshot for the run (planning / cost layer, not the backtest winners list):
Total bots: 6
Total contracts: 17
Total starting capital: $100,000
Total margin required: $28,000
Total round-trip costs: $28.50
Average (planning) Sharpe: 1.52
Average (planning) Sortino: 1.90
Average (planning) win rate: 53.0%
Average (planning) max drawdown: 15.0%
Those averages are the design targets in the portfolio workbook. The backtests told a harsher story.
The profitable-bots companion report for the same folder is blunt:
Profitable bots: 2 | Combined P&L: $31,794.46 | Allocation: $50,000 | Portfolio return: +63.6%
Again: simulated. Educational. Not a live track record.
The top performers (by backtested net P&L)
Here are the only two names the pipeline labeled profitable, with numbers as stated in the run’s profitable-bots / backtest outputs:
RankBotSymbolBacktested net P&LWin rateSharpeProfit factorMax drawdownTrades1Micro Gold Safe-Haven MomentumMGC (COMEX)$25,352.5051%2.121.7976.3%672Micro 10-Year Treasury Yield Curve SteepenerZN (CBOT)$6,441.9648%0.841.2874.7%52
Combined profitable P&L in that report: $31,794.46.
Why gold stood out
The gold bot is the clear leader on net P&L, Sharpe (2.12), and profit factor (1.79). That lines up with the morning report’s safe-haven / real-yield / geopolitical hedging theme: when the narrative layer keeps pointing at gold as a hedge sleeve, a micro-gold momentum expression is exactly the kind of candidate you want the generator to propose.
But look at the drawdown: 76.3%.
That single number is why I refuse to slap “best bot” on a blog graphic and walk away. The same rule engine that likes a profit factor above 1.5 also targets max drawdown under 20%. Gold made money in the sim — and violated the drawdown rule.
So the honest ranking is:
Best P&L / Sharpe / PF in this run: MGC
Still not a clean pass on risk protocol
If you only optimize for “most lucrative,” you will fall in love with strategies that can also dig a hole you cannot emotionally or financially sit through.
Why the Treasury steepener came second
ZN printed $6,441.96 on 52 trades, with a 48% win rate. That is fine if average winners are large enough — and the report’s rates section was heavy on curve / Fed-path expression — but the risk sheet is mixed:
Sharpe 0.84 (below the run’s own > 1.0 Sharpe target)
Profit factor 1.28 (below the > 1.5 target)
Max drawdown 74.7% (again far above the < 20% target)
So: profitable on net P&L, not a clean institutional-style pass. Second place on money; still a research candidate, not a “deploy and forget” sleeve.
What did not make the lucrative cut (and why that matters)
Four bots did not show up as profitable in the same filter:
MCL (Micro WTI geopolitical hedge): backtest marked not profitable (negative net P&L in the bot plan metrics).
MES (Fed pivot equity-index momentum): backtest marked not profitable, with severe drawdown vs the 20% target.
MBT (Micro Bitcoin regulatory hedge): backtest outputs in this run look pathological on drawdown / P&L scale — treat as a failed / invalid sim, not a tradable result.
NG (Natural gas seasonal spread): 0 trades in the backtest metrics (and a contract entitlement / access issue in the plan), so there is no P&L story to tell.
This is the part most “AI trading” content skips.
A news-to-bot pipeline will happily invent six stories. A backtest with rules will throw most of them away. That rejection step is the product.
Also notice the gap between planning expectations and realized sims. The portfolio sheet floated expected-return ranges like $3,000–$12,000 for oil or $2,400–$9,600 for MES, with planning Sharpes around 1.4–1.7. The backtests did not rubber-stamp those hopes. If you only publish the expected column, you are marketing. If you publish both columns, you are teaching.
Risks I want you to keep in front of you
Drawdown is not a footnote. Both “winners” printed ~75%+ max drawdown in sim. That fails the run’s own < 20% drawdown target.
Sample size is modest. 67 and 52 trades is enough to screen, not enough to crown a forever edge.
Overfitting / narrative fit. When bots are generated from the same day’s news thesis, they can look clever for reasons that will not repeat next week.
Backtest ≠ live. Smoke tests, data entitlements, and broker/market-data login failures showed up in the same folder. Live microstructure, margin, and fills will differ.
Projected metrics ≠ backtested metrics. The workbook averages (Sharpe 1.52, max DD 15%, win rate 53%) are not what the profitable filter proved out.
Correlation risk still exists. The morning report itself hammers cross-asset correlation rules. A gold winner and a rates winner are not a diversified “set and forget” portfolio just because both printed green P&L in one folder.
Takeaways for traders and developers
If you trade: treat automated research as a hypothesis factory. Demand the ugly sheet — drawdown, trade count, profit factor, rule violations — before you care about the headline P&L.
If you build: separate layers cleanly:
News / macro brief (what institutions appear to be positioning for)
Strategy generation (bots with explicit symbols, size, costs)
Hard backtest + rule gates (profitability is necessary, not sufficient)
Human review (especially when sim drawdowns explode or a bot never trades)
If you teach or sell tools: show the rejects. Today’s honest scoreboard is 2 profitable / 6 generated, with both survivors failing drawdown discipline. That is more useful than a highlight reel.
Where I am taking this next
I am collecting ideas for AI workshops built around workflows like this — news intake, agent/IDE tooling, bot generation, backtest hygiene, and how to read a profitable-vs-failed report without lying to yourself. If there is a module you want (Cursor/Grok-style agent workflows, Python bot pipelines, dashboarding before the cash open, risk-rule design), tell me what would actually help you.
While QuantLabsNet.com is shut down, the work lives in two places now:
hftcode.com — coding tools, learning, and community
theorderbookedge.com — deep market analysis
And as usual, I will dig into topics like this on my Tuesday night YouTube live stream at 7 PM EST: youtube.com/quantlabs
Bring questions. Bring skepticism. Especially bring skepticism.
Because the most lucrative chart in today’s folder was not “AI found free money.”
It was gold printing ~$25k simulated P&L with a 76% drawdown — and a pipeline honest enough to also show you the four ideas that did not make it.
That is the edge I care about: process over mythology.
Educational content only. Simulated results. Not financial advice. Futures and options can result in losses exceeding your initial outlay.