Risk Takers Podcast - Ep 137 - Rufus Peabody

Podcast Release Date: 2026-01-07
Podcast Link: https://open.spotify.com/episode/1cB5THxJzb3zWflH9c6UxO
hagrinBPR Review Date: 2026-03-03
Rating: 8.25

Notes:
Bias - I find the solo, longform Rufus pod stuff very good and generally much better than the BTP content so going in pretty excited especially since the prediction market stuff has taken off. Not a knock on BTP - different format, different length, just find this format for Rufus when something major has changed / new larger story drops. I haven't reviewed the Risk of Ruin pods officially, but my guess is those will be a slight tick higher than this one due to halfkelly's pristine production.

4:00 - Early Days - LVSC (Las Vegas Sports Consultants), no mobile books, would have to drive around LV betting stuff + offshores (Fall 2008). Big edges at that time - preseason round robins and teasers, 1H, 1Q NFL and NCAAF. Couldn't use a computer or phone in the sportsbook (just a different era). Had to hope the lines wouldn't change as he commuted from book to book. LVSC internship betting MLB correlated parlays - parlaying home underdog with the under. Mentions that an early edge was just having the data through LVSC (again different times, getting datasets back then, even years later was still difficult). 8:00 - Starts working with a group Zach White, Mark DeRosa (EdTeach) - have 4 full-time runners. 11:45 - mentions almost more honor in betting around that time, no real limiting / booting, "took a bet, moved the number" (mentions how Stations would limit you to 1 trip to the counter / day). Enjoyed the sports betting industry 15 years ago more than today (for me it would be more like 10-12 years ago, hard agree, don't love the mainstreamification / legalization, etc).

17:00 - Leaving that 1st Group / Going Out on Own - 2009-2016 with the group, finally splits off. Was doing 90-95% of the modeling, was also sourcing accounts, but at the time still only getting 25% of the profit. Late 2025 gets a call from Mickey Richardson (BetCris). Offered way more money than he was making, but still a hard conversation with the group. Tells a funny story how Train bring on a friend as a trader in a new arrangement and that new guy gets into a love triangle with a higher up at Cris where a bunch of their money is (lol).

Quick side note - I quickly searched Mickey's name because I wasn't sure how it was spelled. I ran across this article from 2003 (before my time) - https://scottberinato.com/bookmaker-whiz-kid/ about fending off a DDoS attack and extortion attempt. Absolutely awesome read since I didn't know this story (and yet another example of how things were so different back then).

22:00 - SP brings up an interesting point about Rufus' longevity in modeling / originating. A lot of guys are more edge seeking / APs where as Rufus has been modeling / originating for a long time now, not a ton of guys with that longevity and then still that amount of success (which makes the PMs so valuable here, you can tangibly see the success (yes, yes only part of the picture I know, but better than nothing)). Mentions Train gets a golf dataset, Train betting H2Hs, Rufus takes the dataset, figures out how to time decay the skills and just starts producing numbers and betting immediately. Confidence to do so because of Train's H2H success.

28:30 - Exploring New Sports - "Haven't dove into new sports in years." Talks more about small edges, being creative, logical (all very good advice obv). 30:00 - "Jeff always says you should understand where your alpha is coming from", but Rufus explains sometimes that's very hard to explain why there's an edge on one event vs another and then verbalizing that narrative. GP reacts sort of how I did initially that "you could (should?)" be able to. (This is a pretty interesting discussion, thinking about my own bets it's often framed as "lower/higher than market here because of x I think", but it's not always easy to elaborate). 31:45 - Mentions ramping up NCAAM is all Telemachus (smart guy, hope we get another pod with him soon), builds out a simulator which allowed Rufus to feed in projections. Meet by Telemachus writing an article for Unabated and payment was an hour of Rufus' time to talk about data science / baseball. 34:30 - "Can't scale golf in the same way" like they could with NCAAM and NCAAF. 36:45 - "I met Mark following him out to a casino parking lot he almost maced me because I am a very intimidating guy" (lol) and they gave me a 20% freeroll from the get go. "How much do I owe them and for how long?" "I'm not an indentured servant, but it had been long enough / the deal was no longer fair". Good discussion about partnerships, percentages and Rufus inferring this idea about loyalty. Mentions they review their splits once a year (which seems very smart). Fairness, trust, respect all important attributes with partners - also mentions you have to like working with them.

43:15 - MrPeanutBettor question - "How much of the edge is team evaluations vs situation / spot?" Rufus says he thinks its 95% evaluating teams. Understanding the dynamic nature in team quality "you want to have less lag in those changes in team strength than the market, you want to get there first" (I like how he expressed this). Learned from Kenny White, his boss at LVSC who built power ratings. Has tried quantifying things like "let down spots", but mentions the difficulty behind doing this. 48:30 - SP follows this up with a really smart follow up question - he pictures having better power ratings as a) better data b) better contextualizing c) model choice / recipe. (I think this is a great way of breaking this down). Rufus - obviously, you want good data, you don't want to be at a data disadvantage, but not the main source of alpha. 53:20 - "There's a 4th thing - asking the right question of the data." Says his strength is finding fixes / hacks / "kludges" (my word) to questions asked of the data vs having this extensive textbook statistical knowledge.

59:00 - Machine Learning / AI - GP keeps using the term "black box" and I think we're in Princess Bride territory. Rufus says he's not smart enough to use ML or AI, I would disagree. "I have been told things I have done are machine learning" followed by "all in the regression family".

1:01:15 - Prediction Markets - First started seeding some golf stuff. 1:02:40 - says something very interesting here that processing last week's data takes about 8 hours. This is something I struggle with too and I used to compliment one of the football sharps I spoke to with how fast they were able to turn around the last week's data to be ready for numbers in the market. Seeing the liquidity on college football was the trigger for getting more involved in PMs. "Just assumed SIG was perfect, but everyone makes mistakes." "If you're just top down it's harder than if you're seeding with an opinion." 1:06:00 - Thinks he can edge stack with the derivative tools he's built out. 1:07:30 - Mentions G Street Analytics which was going to be a soccer betting hedge fund. 1:09:15 - Excited by market making on PMs, reverse engineering bot behavior, etc. Mentions he's never really been a "trader" which is an interesting perspective. Discusses queue logic, disguising your order, etc. I hate this "passive market maker" term that's being used - it's maker / taker, no need to reinvent the wheel here. Both are crazy successful, but it's funny coming from basically a 24/7 world for 11 years and listening to this section because the lingo is a tad off, but the concepts are obviously 100% since they crush. 1:17:15 - Immediately after this is an excellent example from Rufus using Justin Rose futures as a real world example of how you would attack exchanges / PMs but also in a way that would disguise your position / not leak information (probably one of the best moments of this pod, which is filled with great moments, Rufus probably saying too much here). 1:21:00 - "Market making is all games - there's just different levels to it". Figuring out what a bot is doing, figure out its logic and how to exploit it (its a shame Rufus never got into crypto, he would have loved Flashbots, sandwich attacks, etc. If you think this is fun, that era was like the best video game around). "A good bot should have randomness." (This is a great idea and thought exercise - how would you introduce valuable randomness into your bot?). 1:23:40 - Rufus then says something I don't agree with fully - "you're never going to have a bot that's better than a really good trader". My experience is this just isn't true with caveats being - 1) certain markets and market conditions, bots will always win because humans quite literally cannot act fast enough, 2) it's more prevalent in financial markets than sports markets at the moment, but 3) as t time increases and assuming prediction markets are allowed to operate in manners similar today, the gap will narrow for even markets and market conditions that don't require to be "first". Again, "be first, be smarter or cheat". Humans are absolutely much better at #2 and more prevalent in sports, but the use of superlative "never" is just wrong. Rufus follows up with bots and their ability to scale an operation beyond what a human trader can do. 1:26:30 - GP mentions that prediction markets are so new. I hear this a lot - there are aspects that are new, there are aspects that are new to the very sharp guys who were winning at sports, the structure of these markets is not that new. This goes all the way back to the 2004 Levitt paper on sports vs financial markets and if you've participated in both and understand the differences, there's really not a ton "new" here. I think you should be careful conflating "edges in PMs" especially sports PMs of which there are obviously new edges all over vs "futures contract markets are new". Again, both guys crush and understand the concepts super well, just some of the jargon needs to be fleshed out I think. (I also think I am spoiled - I've been friends with someone who is MMing many venues now for over a decade via crypto/sports and I think I've just absorbed a ton from him in the last 10 years). 1:29:00 - Talking about "if your bot takes" and both go absolutely not. Did no one bot before PMs on this pod? Some of the issues their talking about, I ran into almost a decade ago now with my most infamous example being the year Lebron was on the Heat and even the in-arena official NBA data feed was broken and so the score froze through all data feeds, yet time kept ticking away before I had to hard code better risk exposure limits logic, but this was almost 12 years ago I think. I really think both groups would probably find great benefit in hiring a high level real-time trading dev (think RenTec) and don't just vibe code everything and a lot of these very basic bug examples aren't a thing. I think this is the limitation of non-coders having this discussion that haven't coded at "stakes" before.

(This was the best section of the pod, just because I have minor quibbles doesn't mean I didn't think it was great - it was great. You should be able to ask and answer all these questions and do deep dives through all these sections yourself. Great discussion from all 3).

1:33:00 - Questions Section - Porter asked a question that no one really understood and rightfully so. 1:39:00 - Thoughts on the Long / Short article - "Shipper vehemently hated it" (lol). 1:44:00 - How do you define the best originator? Rufus through out a bunch of metrics all of which could be answers. "But who really cares?" Great answer from Rufus. 1:47:10 - Broner question - Regulatory framework - "this question should be its own podcast". Outlaw mobile gaming all together! Make sports betting licenses easier to get. Sports betting needs "friction" because of betting's negative externalities. Rufus then mentions a centralized authority for responsible gaming and this idea makes my skin crawl (basically idPair). 100% against iCasino. Great answer. Rufus then gave a long answer to a question about life and it goes exactly how I would have expected - most people are generally good and then the human race is extinct in 200 years. Good stuff there. 2:05:40 - "Irreducible Uncertainty". 2:06:20 - Hot Take - "CLV isn't as important as people think if you're an originator". Does mention the concept of creating his own CLV in golf (which is an important concept for the legit top bettors).