If We Win, Why Do We Need To Sell Picks?

A Completely Honest Answer

You, or someone you know, has run the standard objection: if you could actually pick winners, you would not need to sell them to me. It sounds airtight, but I'm going to show you with numbers from my own accounts why it is also wrong (or incomplete, at worst). This is what betting with an edge for us really looks like from the inside. Warning: this is long but it’s necessary to understand the full picture.

I am 36, and I spent the first 30 years of my life in school. From 18 to 22, I worked towards my bachelor’s in chemistry. Then a year off teaching at the University of Arizona (Bear Down!). Then five years of graduate school for my PharmD plus an M.S. in pharmaceutical economics, outcomes, and policy. Then a postdoc in health economics and outcomes research (HEOR). At 31 I landed my first full time job in the HEOR field, and today I am a director and it is an accomplishment of which I am proud.

Somewhere in there I got very interested in pointing those methods at sports outcomes. I knew nothing about markets or professional betting, only how to use data to assess outcomes, which was literally what I was being trained to do. After years of tinkering in my free time, by 2022 I believed I had built an NFL model that could win. I was going to beat Vegas.

Instead, I got my shit rocked. Shout out to anyone following us from day one. We lost a lot of weeks. I had no concept of CLV, no idea how the markets functioned, I just knew that this big bad 'Wizard of Oz' was whooping my ass every Sunday. But the problem was too interesting to quit, and eventually, it clicked. I was not betting against a basement full of Vegas geniuses. I was participating in a market. At the time my modeling ran on DVOA, a fully public metric, which meant I was trying to beat the market with information the market already had. That was never going to work. Then someone handed me the bible, The Logic of Sports Betting, and things dramatically changed from there. With help from people like Tej Seth and Alex Laycock, we started over with our own variables and built new models, and now that we knew what to look for (things like CLV) those models could compete in the market. Since we started charging for access to our model plays, we have never had a losing season. That could change, variance is undefeated, but to me the streak is a testament to learning and iterating rather than to luck.

I lay that out not to brag but as context for BTB Analytics and for why the money printing myth falls apart. In 2023 we rolled out the new models and, sure enough, we won. We have won every season since, and so have our clients. So why sell picks at all? Just win and print money, right?

The Numbers

Now for the black box.

In 2023 I had roughly a $12,000 bankroll and was betting $330 per game. We won about 10 units that season, so I won around $3,000. The money printing machine returned $3,000 on a $12,000 roll. As a percentage return that is good. As a making a living from it, it does not even cover one month of my bills. That same year I made $180,000 at my corporate job with the same skill set and the same effort. Honestly, this endeavor took more effort, but the day job paid 60x more.

Fine, you say: just bet more, dumbass. I was right there with you, so in 2024 I did. I bet $660 on every NFL game and $330 on every CFB game where the models showed an edge. And we won again, though not by as much as 2023. Even though I had doubled my sizing, I made about the same $3,000. This time I pushed over $100,000 in volume through the books. Week 1 of the NFL, stacked on Week 2 of CFB, I had just under $8,000 in play. Across 2024 we averaged $4,000 to $6,000 in action each weekend.

Then Week 4 arrived. Bam, we started getting limited. BetRivers first, then DraftKings, and others. I wanted to bet more and ran headfirst into a problem that wasn't obvious before: if you have an edge, the recreational sportsbooks do not want your action.

The bigger cost was the one rarely anyone talks about. Having $4,000 to $6,000 riding every weekend was gut wrenching for me. Wins did not feel like wins. They felt like relief that I had not lost. Losses felt like getting punched in the face. My relationships suffered, my mental health declined, and I still had to step back up each week and keep firing as variance hit. All of that for a ~$3,000 return, in a year when a promotion took my day job income to $240,000.

In 2025 my day job paid me $240,000 again but I kept pouring effort and resources into the models, on the theory that the more I understood the models and what each week feels like, the easier the sizing would get. I improved the CFB model. The result was our most profitable season ever, more than 30 units won across CFB and NFL combined. I took the lesson of 2024 and went back to $330 on every game, roughly $3,000 a week in action, with some weeks up in the $6,000 range

Betting Big Is Its Own Skill

Betting large sums of money is a skill in its own right, separate from having an edge, and it is one I do not have. I would put real money on the claim that nobody thinks harder about extracting an edge from our models and data than I do. Yet betting thousands of dollars on each game is something I cannot do well. Some weeks I can pull the trigger. But it takes a toll, and the toll arrives at the worst possible moment, right after a losing weekend, which is when discipline demands you step back up to fire on your edges regardless if you are down on the season. I do not consistently have that in me, and I do not want to pay the pain it costs to build a skill I lack.

What we do is simple at its core: find games with edges, bet them in a disciplined, planned way, days before the market closes, measure CLV, and make money. For me, and I suspect for most of you, loss aversion is real. Whether your unit is $100 or $100,000, losses probably cost you more emotionally than wins pay you. They certainly do for me.

I have written elsewhere that the skills powering all of this are very employable. I included my salary figures, at the risk of offending, because they are central to the calculation. I am not betting to feed myself. I am betting because I have an edge. Betting large is not necessary for me to make a living, and what is not necessary is much harder to justify paying for in pain. Loss aversion means I cannot bet the sums required to replace my salary. Period.

So Where Does The Money Printer Live?

Maybe more important than whether you are psychologically capable of betting large is whether the market will even let you. Who is willing to take big bets from anyone who wants to make them? As we have covered, it is not your favorite retail book. I, like many others, got limited by multiple books while betting around $500 a game.

So let's do the math on replacing a salary like mine. Say the target is $200,000 a year. We assume, conservatively, a 3% edge on CFB and NFL. We bet about 75 NFL games and 100 CFB games (at open) a season, and we get most of them at an average price of -110, which carries a break even of 52.4%. Of the ~175 total bets, assume we win 96 and lose 79. How much per game does it take to clear $200,000 a year? The answer is an average of about $24,000 per game, which works out to roughly $4.2 million in total volume across the season. If we push the hypothetical win rate all the way to 58% (highly unlikely, btw), and the required bet only drops to about $10,700 a game, still around $1.9 million in volume.

Now layer in how you are supposed to size these. Kelly, at the fractional level any sane person uses, puts a single wager at roughly 2% to 3% of bankroll for an edge like this. So a $24,000 average bet is not something you fire off a $50,000 roll. It implies a bankroll of $800,000 to $1.2 million just to size responsibly and survive the variance. Want to be a real syndicate and fire $50,000 a game? At 3% of bankroll, that is $1.6 million sitting in accounts behind you. The bet and the bankroll it demands scale together.

Even with the stomach and the bankroll, you cannot get that size down at the number you want. At market open, when the edge is biggest and the CLV is best, you can usually max bet about $5,000 on a game, often less. There is no world where you drop $24,000, let alone $50,000, into a fresh, soft number. You wait for limits to come up and take a worse price, or spread it across a dozen books and beards while chasing the line as it moves against you. Either way you are bleeding out edge. So no, we cannot bet that large into a good number, anywhere, ever.

So where does the money printer actually live? In my case that is a corporate job, applying similar methods to problems other people consider more important, or have the capital to spend on. It is not in becoming a sports betting syndicate where I would need to fire $24,000 a game every weekend, losing myself and likely my family and friends in the process, to make less money than I already do.

Which brings us back to why we sell picks. My nature makes me bad at betting the large sums required to extract large sums from sports betting. Just as important, the same traits that let me find an edge also mean my opportunity cost is high. I can make the same money or more, with exponentially less risk, doing other things. That is where BTB Analytics sits today. We can try to scale this business by helping people find edges and use them to make real extra money. If you are unlike like me and are comfortable risking $20,000 to $30,000, you can see nice returns in short windows. I do what I love, hunting for edges and understanding why they work, I share it with people who want to learn, and I avoid the one thing I am bad at: betting large.

So does the money printer exist? Maybe, for a select few. The Rufus Peabodys of the world have the knowledge to find an edge and the temperament and bankroll to fire $50,000+ a game all season. Even then, there are far more people like me than there are people like Rufus. People with the skills and knowledge to compete in the market and win, but without the stomach it takes to truly capitalize on it and replace a full time job.

That is where we sit today. Can we scale BTB, get enough users to replace our jobs (admittedly a high bar), by growing a community who want sharp bets and want to learn? If so, we keep building and make more models. If not, we shut it down and bet the models for fun, just at more normal amounts. The money printer, for us, lives in the scale of who else we can help, not in unlimited money extracted from the market.