LCSLarry Esports Model 2025 Results: +1,761 Units, 30% ROI, and What Actually Happened

2025 is in the books, and by any objective measure, it was a fantastic year.

That's real edge, at real volume, sustained over 10 months. But if you tailed Larry this year, you know the numbers don't tell the whole story. The year wasn't a smooth ride to +1,761 units. It proved why volume matters, why variance humbles everyone, and why most people still lose even with an edge.

Let me walk you through what actually happened.

The Monthly Breakdown

Look at that table. Every month was profitable except one. And yet, I guarantee you, there were moments this year where people in the Discord or on Twitter were convinced the model was cooked.

That's betting. Let me explain.

March to May: The Dream Start

March was our first full month of tracked results, and it set the tone: 63% hit rate, +55% ROI.

April topped it. 66% hit rate, +88% ROI. Our best month of the year. From March 30th to April 27th, we had 28 winning days and a single losing day.

By the end of May we were up +755 units. Everything felt inevitable.

Then summer arrived.

June & July: The Low Volume Months

June and July were rough. Not because the model stopped working (June had a 44% ROI on opening lines) but because there was nothing to bet on.

No CS2 model yet. League was in the MSI dead zone. We had one Dota tournament and one COD tournament running at a time. Volume dropped to 700 bets a month, down from 1,700 in May.

When you're only placing ~6 parlays a day across a handful of matches, everything is correlated. One upset ruins your slip. Variance hits harder. Even if the edge is there, you don't see it.

July finished at 7% ROI. Technically profitable. But the vibes were terrible.

August Changed Everything

We shipped the CS2 model in August.

Overnight, volume exploded. We went from 700 bets in July to over 4,100 in August. The ROI jumped back to 29%.

This is the thing people don't internalize: volume is the product. When you have an edge, more bets means more opportunities for that edge to compound. It means less correlation between your positions. It means variance smooths out faster.

August proved it. Same model philosophy, same discipline, just more surface area.

September: Slow Start, Strong Finish

September started slow. The first two weeks, we were treading water. Sentiment in the Discord was negative. People were questioning whether the CS2 model actually worked or if August was a fluke.

Then September 23rd hit.

The mood flipped instantly. Everyone was celebrating. "Infinite money printer." You know how it goes.

October: Our Biggest Month

The heater lasted until October 10th.

The next week we added only 47 units. Still profitable, but it didn't feel that way. People who sized up during the streak got overconfident and stopped managing risk, expecting every day to be green. When the correction came, they were overexposed.

October finished at 4,500 bets and 29.5% ROI, our best month of the year. Most of those gains came in the first ten days riding the tail end of the heater. The rest of the month was steadier but less explosive.

November: The Hot Streak Returns

November started choppy. Overall volume for the month dropped to 2,800 bets, which didn't help. For the first two weeks, we pretty much broke even, down about 5 units by the 14th. Individual results varied: some people were green, others were red. Sentiment was back in the gutter. Same story, different month.

Then November 14th: another heater.

From November 14th to the 28th, we went 808 wins to 471 losses, a 63% win rate over two weeks. We ran up approximately +200 units and had maybe one red day the entire stretch depending on your book.

By Thanksgiving, euphoria was in the air. November finished with a 59.4% win rate and +24% ROI.

You can probably guess what happened next.

December: The Off-Season

December was the first red month in Larry history. Across all books, aggregated: -11% ROI.

It was also the lowest volume month since we added CS2. Only 1,400 bets. League was off. Holidays killed the schedule. We were betting on a handful of smaller CS2 events and one Dota tournament.

Everything that made summer rough came back. Low volume. High correlation. Sharper lines because the books had fewer games to worry about. Lines got bumped faster because all your bets were hitting the same few matches everyone else was betting.

The model didn't break. The conditions did.

But here's the thing: if you zoomed out and looked at the year, you'd see a 30% ROI across 23,000 bets. December was noise. It just didn't feel like noise when you were living through it.

The Actual Lessons from 2025

1. Volume is everything.

Every period where sentiment was bad (June, July, and December) was a low volume period. When there's not enough to bet on, variance dominates. When volume is high, the edge asserts itself.

2. Hot streaks end. Cold streaks end.

September 23rd to October 10th: +323 units. Then a correction. November 14th to 28th: +200 units. Then December happened.

Variance always reverts. If you size up during heaters expecting them to last forever, you will get humbled. If you quit during cold stretches, you miss the inevitable bounce.

3. Risk management is the product.

The model did its job in 2025. It found edge. It printed 1,761 units. But individual results varied wildly based on how people sized, when they started, and whether they stayed consistent.

The people who did best were the ones who:

• Sized conservatively
• Didn't chase after hot streaks
• Didn't quit during cold streaks
• Understood that any single day, week, or month means nothing

What's Coming in 2026

January is prime season. League regular season is back with 6-8 leagues running simultaneously. Dota and CS2 tournaments are stacking up. COD is returning. Volume is about to explode.

Here's what's shipping:

• Valorant model: Testing is done. Launching Q1.
• Dota model update: Significant improvements, shipping soon.
• COD and League updates: In the pipeline.

The foundation from 2025 is solid. Over 23,000 bets of data. A proven edge across multiple esports. And now, more games and better models.

If you learned anything from 2025, it's this: stay consistent, manage your risk, and let volume do the work.