Bet the Process - Matthew Trenhaile

Podcast Release Date:
Podcast Link:
hagrinBPR Review Date: 2024-08-01 08:59:12
Rating: 7.75

Notes:
Bias - So my bias going into this one is that by the end of this podcast I would have listened to like 7 hours of Trenhaile in 2 days. I'm not sure I could listen to anyone for 7 hours talk about anything without fading in and out of consciousness (which I am fully aware I once talked for 4 hours and 10 minutes and the host literally started falling asleep). Quite the media blitz from the 'ol chap.

Starts with some background info. They ask Trench "why listen to you" which I thought he answers very well. (Conversely, no one should care what I think, these reviews are a way to quickly digest sports betting content).

4:30: Bearding / Multi-Accounting: Rufus bring up the Schwimmer tout idea *I think* in earnest and I about had a stroke. Trench does make the salient point that if the regulators/states allow the operators to define these terms in their T&C, then they will shape those definitions / "tool kit as they see fit". Funny enough Jeff interjects that instead of the legal arguments, he wants to know where Trench draws the line. (Before I press play, I would say opening up bank accounts and carding stemming from using other people's IDs probably the wrong side of the line, but let's here what Matt says). "Profiling is the cornerstone of all risk management." We take a windy, quick path down "the ethics in capitalism" and we end up talking about the benefits of profiling to the overall customer base by weeding out the "less valuable customers".

13:00 - Player Profiling: "Winning bettors by definition are outliers." Straight bets, timing of bet placement to start of event, etc. (stuff we've heard many times before). Brings up method of deposit back in the old days (Neteller) and how squares weren't using e-wallets, depositing by debit card was easy. (Maybe this was different in the UK, but I don't remember this being remotely true in the early 2000s, I used Neteller because I had to.). Mentions profiling based on zip code (think some delivery companies have tried this and have found themselves in some very uncomfortable social equity situations). 16:30 - First really interesting point - "but you can predict customers before they've placed a bet with a certain degree of accuracy". (This is the thing that freaks me out about things like idPair which is the worst version of this topic.). 23:00 - Rufus brings up this concept Trench brought up on another podcast about "subsets of bets that only look square". Trench gives the very smart "well you have this Unabated data" remark and then ties it into very popular player overs only props. Then, he does it again when he says "You know if you query the Unabated data and you see 80% of the value were on the under, you found the bookmaker bias" which is clearly the value prop in building tools like this (they are obvious, but having the data to show you exactly how biased these markets are is extremely helpful).

26:30 - Hard to Model/Price Events: Fight/combat sports. 30:20 - Brings up talking to very smart people who couldn't come up with consensus on Mayweather/McGregor - "I don't know is there a knockout risk here?" "Best bet of the millenia didn't fully set in until afterwards." /blinking_guy_meme.gif. Election betting.

34:00 - Betting Career: Trench brings up a point that I don't think I have thought about much. The new bettor that builds a model and then just starts betting a model, that others with an extremely similar model have had great success with, but just due to timing they start betting during a period of bad variance and they give up after a few months. Not sure I've ever thought about that exact scenario, pretty wild when you think about it. "Do you have the almost zero emotional intelligence" (I feel seen.) "Fighting against tinkering." 40:00 - Talk about the lifespan of edges / influence of model in the market, Trench brings up some fun old school history with spread betting at IG Index where they had a golf simulator running Monte Carlos that used Jeff Sagarin's ratings. 42:30 - Mentions that early DataGolf, the further you moved away from the outright winner markets to the make/miss cut markets, the higher the ROI was if you followed the early DG numbers (interesting, not a golf guy didn't know that). 51:00 - Quick Dutching mention - I actually didn't know what this was formally called until only just recently, but Dutching is a concept that everyone understands intuitively, but you'll find different applications of it (arbitrage, profit locking on an event, bets to group outcomes/sides on a single market, etc) & Trench mentions there's software out there that can expedite such a process.

Whole bunch of talk about what rec bettors do. Just pull the George Costanza here - "I'm unemployed and I live with my parents." Just take your every instinct and then imagine the opposite to figure out what the rec bettor thinks. Jeff then asks about election betting and Trench says to almost do like a polling of people around you ("talk to your incredible biased father") and this is horrific advice we're starting to lose the plot here. "You need to find alternative data sources like Facebook Groups." Did someone just body snatch Trench, what is happening? In fairness, he gets the train back on the tracks, mentions looking at it from a bottom-up approach, but that was quite the detour. 1:09:55 - The election betting stuff finally ends.

1:10:45 - Rec Books Moving on Air: Think the most interesting point in this section, maybe in the entire pod for me, was Trench asking Rufus "how long can a line stay manipulated?" I actually don't know the answer to this question & this was a lively back and forth. Rufus gives an example he's made before on a podcast about manipulation either in the same direction as DataGolf or differing & the length that manipulation will survive will last differing amounts of time. Trench fires back that it's "utterly obvious to the operator" when a line is being manipulated. Interestingly enough, Rufus makes the example if he places 10 bets at Circa, 9 will be legit, 1 dummy to which at 1:14:50 Trench says "your account could go for 10k bets and people could just assume that you're playing straight" which is quite the statement after "utterly obvious". 1:16:30 - "What would happen if these price setting books disappeared tomorrow? People still know how to price things." (We sure about this? When the Costa Rican trunk cable was under maintenance pre-SpankOdds, it was Thunderdome pricing wise especially for things like 2Hs. I think Matt's point is on a longer time horizon than some emergency undersea cable maintenance outage and also at larger legal operators & there I agree.) This for me was the best 12 minutes of the podcast by far.

1:23:00 - Rec book model talk. Trench argues the model is many millennia old, some interesting history of ancient Egyptians with weighted dice, horse racing, etc. This transitions into some discussion about Unabated and the tools offered and some light touching on scaling and what impact tools like say the NFL Simulator have on a market. Gets back to the rec book talk and the "one directional nature of costs with vices" which was discussed on another Trench pod. (I actually think the most compelling thing Trench has said on this topic was in the CO pod where ops are calculating the compute costs of a user especially in the day of these live & SGP trading engines, cash out pricing, live bet tracking, parlay leg values, etc.)

1:43:00 - Lessons from Bookmaking for Betting: "Persistence." "Seeing what has worked for people, their bets, their consistency." 1:47:30 - Rufus brings up the CO comment about "successful people and overbetting" and Rufus brings up the fact that this could very well be survivorship bias. Trench counters with a lot of these people might not even know they are overbetting at the beginning & only figure this out later. (This still sounds like survivorship bias to me, I think Rufus has a solid point here.) There is a good point made about lifetime risk vs the % risk at that moment in time (esp young people).

Summary - Always great, think both this, the Deep Dive and the CO pods were all great. The CO pod was just a notch above as I found that one more logically structured, this one was more a free flowing convo between smart people with slightly less structure. The 1:10:45 section is maybe the best section though of all the Trenhaile pods as it has made me think and continues to make me think what the right answer to that Q is.