March 2026 Recap: +257 Units, 25.7% ROI
4,003 bets. Three straight months of green. Thunderpick went live. Plus a new LOL model already cooking in April.
+256.72 units. 4,003 bets. 25.7% ROI.
That's March 2026. A $10 flat bettor following every play cleared $2,570 in 31 days. Three months into 2026 and every single month has ended green. Running total: +760 units.
February ran 31.5% ROI and I said at the time it wouldn't repeat every month. March at +25.7% is closer to what a sustained month looks like for this model. Still enormous. And it came alongside the biggest new release of the year so far: Thunderpick straight betting went live in full. More on that below.
No single game carried the month. VAL hit 66%. DOTA cooled off. CS pumped out 2,500+ bets on its own. Average it all and you get +25.7%. That's the whole point of betting five esports in parallel — when one cools, another goes off, and the portfolio stays green.
CS2 is the workhorse
1,486–1,021 at 59.3% win rate, over 2,500 bets on the month. Over half of March's profit came from CS alone. No other sport contributes volume like CS does right now, and the raw bet count is what makes the compounding happen.
The rebuild that went live February 13th is doing exactly what it was built to do — grind out +EV plays at scale, day after day, and hold a win rate well clear of the breakeven line.
VALORANT is still the king
Two and a half months after it launched, I was expecting the VAL market to be sharpening up. Books adjust, samples regress, edge dulls. Every new market follows that arc.
VAL hasn't gotten the memo.
114–58 for a 66.1% win rate. February was 61%. The number went up, not down. Books are still sleeping on VAL props. I don't know how long that lasts, but while it does: bet every one the model spits out.
LOL stays dialed (and there's a new one)
181–108, 62.7% win rate. League did what League does — still the most dialed model on a per-bet basis, still 60%+ month after month, still the foundation everything else gets built on top of.
One thing to flag: I shipped a new LOL model in April. Early numbers are even better than what you see above. Yes, this recap is late — I was busy building the next thing. Full breakdown in the April recap.
COD keeps printing
COD launched February 19th. First full month on the books: 304–200 at 60.4% win rate. Second-highest profit contributor of any sport this month, behind only CS.
Every new model here has followed the same curve. The edge is biggest early because the market hasn't priced it in yet. Ride it while it's there.
DOTA 2 cooled off
305–226, 57.5% win rate. DOTA was the model's hottest sport in February. March regressed — same model, same pipeline, five points of win rate gone. That's variance on a sample this size, nothing more, nothing less.
This is exactly why the thesis is “bet every +EV spot across five esports.” In a month where DOTA cools, VAL goes off. Two months from now it'll probably flip. You can't predict which sport gets the heater — you just have to have exposure everywhere and let the math average out. Which, for March, it did.
DOTA still ended the month green. Just quietly.
Thunderpick: the numbers are in
Thunderpick went live on the model in early February. The Feb recap teased it with preliminary numbers. There's enough data now to stop calling them preliminary.
All-time, across every straight bet the model has flagged on Thunderpick:
- Record: 6,898–5,097
- Win rate: 57.5%
- Net units: +540.78
- ROI: +4.5%
Twelve thousand graded straight bets. +540 units of profit at a real sportsbook, not DFS parlay math. Pro sports bettors spend years trying to hit +3% on traditional books and mostly fail. The model is doing +4.5% on everything, unfiltered.
Apply any discipline on top — filtering to higher-EV plays only — and the returns explode. At a +10% EV filter, the same sample becomes 2,117–1,473 (59.0% win rate), +376 units, +10.49% ROI. Per sport at that same filter: VAL +15.28%, LOL +14.42%, CS +10.77%.
Those numbers are fake-looking high. They will almost certainly regress as books tighten and volume grows. But right now it's what the data says — and the trend has been up, not down, as more data comes in. The model ports to straight betting as cleanly as it runs on DFS.
Thunderpick doesn't want to work with us anymore. They pulled our affiliate deal at the start of April. Stated reason: they were losing too much money on our users. We used to get paid per signup. We don't anymore. Because the signups we sent them kept walking off with the house's money.
That's the cleanest endorsement of the model I can give you. When a sportsbook fires you as an affiliate because your traffic is too profitable — for the traffic, not for them — the edge is real. Pour one out for the CPA check. It died doing what it loved.
What's next
April is already in motion:
- New LOL model is live. Early numbers are absurd. Full breakdown in next month's recap.
- Bovada and Stake both went live on the platform at the start of April. Two more real sportsbooks in the mix, both with straight-bet esports prop markets.
- Polymarket is already being scraped for team and game-scope esports props. Straight-betting model on top of it is next. Prediction-market liquidity, esports-specific markets, no DFS math.
If you want to try it, there's a free 3-day trial. Five esports, twelve sportsbooks, scanning prop markets 24/7. Takes 15 minutes a day once you get the hang of it. See you in the April recap.
