ᴛʀᴀᴅᴇʀᴄᴀʀᴛᴇʟ™

16% a year for 13 years. A third of the stock market's drawdown. Sharpe 1.5. And it's accelerating. 24% a year since 2022. It's called Gravity Well. A systematic edge I built, run as a full system. Here's how it holds up 🧵

Lets bring back forums...

@KolbeAgency 1-0 Norway win

@ramtrades28 @Topstep Congrats man. First of many. 🔥🔥

My first-ever payout! I've blown a lot of evals and funded accounts. I kept searching for the "perfect" strategy, but eventually realized that everything comes down to risk management in trading. I still battle revenge trading and oversizing, but @Topstep's risk settings (DLL and position size limits) have helped me stay disciplined and in control. Still long way to go.

Coming soon. Gap Stats Pro on NinjaTrader 8. Same engine as the TradingView build, same 24,168-gap validated priors. But NT has no bar budget, so the table will stand on years of your own one minute data instead of leaning on the seed. Native alerts. Prop-preset trade plan. All three sessions. In validation now. Ships when the numbers match. $ES $NQ $YM

@iMATTHEWRYAN Zero-alignment club when? 😎

ChatGPT said that I dislike situations where people implicitly demand alignment, emotional labor, or ideological loyalty lol

Heads up on two bugs. hits / n * 100 is int division in Pine, so FULL% will print 0% on nearly every row (use 100.0). And the CURRENT SESSION cell writes into an already-merged range whenever the gap isn't in the smallest bucket, which throws a runtime error. And the touch/fill flags are only computed vs the 9:30 gap, so all three levels report the same stats. Easy fixes.

@notprofgreen indicator shows same statistic values and percentages for the 3 different opening prices, like if they all had the same opening time, it may be a small bug in the code?

Gap Stats Pro v2.7 is done. The priors got six times deeper. Every cell is now seeded from 24,168 validated gaps, eleven years of ES, NQ and YM one minute history, up from two years. Why it matters. The two year window was a bull sample and it overstated down-gap fill rates by up to 7 points in the bigger buckets. Eleven years washes the drift out. Also in v2.7: XL gaps beyond 0.7x ATR now show their own cohort's history. On RTH they filled about 19 percent of the time. The gap you most want to fade is the one that historically does not come back. The table, the dots and the alerts can no longer disagree. One decision path. Confluence rows need a real sample (n of 10+) to score. Thin slices dim out. A calendar staleness warning, live band numbers on hover, and the data vintage in the header. You should know what a statistic is standing on before you put size behind it. That is the whole product. $ES $NQ $YM

Update. Still waiting for the money to clear. Told me the 6th of July. Nothing yet.

There might be an edge in weather...

Free will is a model with enough parameters to fit your entire past and zero power to predict your future. Overfit, basically.

Luck is just variance you haven't averaged out yet. Skill is what's left when the sample size gets big enough to embarrass you.

@Wealthsimple will Canadian weather stations be on the board at launch, or just Kalshi's US city gauges?

@Wealthsimple brings prediction markets to Canada this summer. Climate contracts, no sports, no elections. Everyone sees boring. I see thin markets and public settlement data. The model is built. The terminal is live. No real dollars until settled trades prove the model is calibrated. Day 1 ready, not day 1 learning.

@sLowi_ @irlfarmer Most likely because public script, closed source code. TV hates that.

@irlfarmer What’s the reason behind them taking it down? If it’s just supposed to be a model idk why TV would take it down

Of course trading view took this Down the next day, I have included the source code and happy independence to us all 🇺🇸

Next test on the bench: Forced Flow. The idea: bursts of forced selling and buying leave a signature. Consecutive directional closes, expanding ranges, volume spikes. The question is whether that signature is tradeable on index futures. Two event definitions. Two hypotheses, continuation and fade. ADX regime gate on, off, and inverted. 12 cells. Full costs. Stability splits. Deflated Sharpe against every trial. The real question isn’t whether the signal prints. It’s whether the gate earns its spot or just deletes trades. If it’s null, we publish the null. Context, not autopilot.

Ruin is the only state you can't recover from. Optimize for staying in the game, not for being right. Survival compounds. Cleverness doesn't.

Tested the V-shape recovery on 11 years of NQ, 543 flushes. Buying the reclaim is statistical zero hiding behind a 64% win rate. Shorting the failed pop is significantly negative. And the prettier the V, the worse it trades. A shape, not an edge. Full study this weekend on NULL RESULT.