Weekly Systematic Trading Recap 02
Let's go!
Happy Sunday, fellow traders.
If your summer has been like mine, you’ve spent half your trading week dodging heatwaves and the other half dodging curvefitting mistakes.
But lucky for you, i listened to hours of podcasts this week so you don’t have to.
This week we’re diving into:
The 4D chess of commodity futures term structure
The real reason your backtest sucks live
Why owning your favorite stock might be dumber than gambling on meme coins
let’s get into it.
Flirting with Models - Benjamin Hoff
Why commodity alpha lives in term structure geometry
Most people treat the crude curve like it’s just a line you roll down. Hoff says, think again. It's a surface and it's screaming signal.
“What connects Dec crude to July crude is the ability to store it. But if you don’t have storage or credit, that connection breaks and the curve decouples.”
“Carry strategies exploit the second derivative of the curve… convexity tells you how much front rolls down relative to back. That spread is your premium for stockout risk.”
My take: He basically turned the crude oil curve into a calculus playground.
Levy area: The alpha no one's looking at
You know all those sloppy lag regressions you tried to model lead-lag? Toss them.
“Levy area captures the oriented area between two asset paths… If one asset tends to move before the other, you get a nonzero Levy area.”
“It’s model-free, time-aware, and captures non-linear lead-lag without making assumptions.”
My take: This is like discovering autocorrelation before the rest of Twitter does. It’s signal geometry. Use it before someone builds a SaaS tool and charges $99/month.
Seasonality isn’t monthly. It’s weekly. And physical.
Everyone (not me) loves seasonal edges. But most of them are fake? Except this:
“Weekly seasonality persists in energy contracts. Independent refiners can’t hedge on weekends, so they rebalance late week based on weather forecasts.”
My take: It’s not seasonality. It’s a predictable hedger flow driven by logistics. The kind of thing your quant friends miss because they never got oil on their shoes.
Line Your Own Pockets - Taking a Backtest Live
From backtest hero to live trading zero
You’ve built 17 filters. You’re still tweaking stop-loss decimals. But you haven’t gone live.
“You get to the point where you do more optimizations and there’s not much improvement. Or to make another big improvement, you're going to have to eliminate a whole bunch of trades.”
“That’s when it’s time to go live. Trade one share. Don’t go for perfection. Go for real feedback.”
My take: Every day you keep tweaking that strategy past version 64final.py , you’re not refining you’re hiding.
Pre-mortem your strategy before it dies live
This one’s gold. Traders do a post-mortem after they blow up. These guys suggest doing it before you trade.
“Try to guess how your strategy will fail in advance. Write it down. If this is going to fail, why? What’s the weak point? Test it now.”
My take: I do this all the time. Step 1: Assume my idea is trash. Step 2: Try to prove otherwise. If it survives, maybe it’s got a shot. If it doesn't? Congrats, you saved 3 months of self-delusion.
Rational Reminder - The Risk of Individual Stocks
Most individual stocks are time bombs
The data here is brutal. You think you're smart because you held $NVDA. But statistically? You're a lottery winner.
“More than half of individual stocks underperform T-bills over their lifetime. 44% of Russell 3000 stocks suffered a 70% drawdown they never recovered from.”
My take: This is like playing Russian roulette with your retirement account. Six chambers, one Nvidia, five Enrons. Good luck.
Even winning stocks tend to underperform next decade
This one hurts more:
“Stocks with top 20% returns over the past 5 years had a median 10-year underperformance of -17.8 percentage points vs the market.”
My take: You fell in love with a high-flyer. But turns out, it's just bloated and exhausted. Past performance isn’t just not indicative it’s predictive in the wrong direction.
That’s the week
What we learned:
Commodity alpha is geometry
Stop optimizing and start trading even if it’s one share
Stock picking is a statistically proven delusion
Question of the week:
How many strategies have you killed too early because they didn’t work live the first week?
Me? Too many. Most weren’t broken they just needed smaller size and less ego.
If you liked this, share it with your trading crew. Or forward it to that one friend who won’t shut up about their 3-stock "diversified" portfolio.