Victory, through value on The Great Triviality

The Journey Ahead

Tue, 07 Jan 2025 11:35:49 +0000

When you set out for Ithaca, ask that your way be long, full of adventure, full of instruction - C.P. Cavafy

The Advent of Alpha posts were a (mostly), fun way to kick things off here, but life got in the way of me doing all the deeper work I wanted, and as a result the quality of the writing isn’t quite where I’d hoped it to be when I started out.

Advent of Alpha Day 25: Shannons Demon

Fri, 20 Dec 2024 19:11:43 +0100

The last day is here. The delivery schedule wasn’t what I hoped, but I’m glad I got 25 posts out. They’re not all great. At least one is related to an idea that one person claimed has made them seven figures. It’s for you to work it out.

I would say the best has been saved to last.

In Fortune’s Forumula, which is a book by William Poundstone that I’ve mentioned before, there is an account of Claude Shannon - definer of Information Theory, the inspirer of Kelly and his criterion - who gave the World a fascinating insight into the mathematics of portfolio rebalancing.

Advent of Alpha Day 24: Order Execution Latency

Fri, 20 Dec 2024 19:11:31 +0100

About six months ago I was watching some youtube videos about HFT trading, looking for ideas, and some guy popped up and talked about a patent for a technique that NASDAQ and other exchanges had to explicitly ban,

The patent basically boiled down to this: you might want to long or short something very quickly, based on a new piece of information. Let’s suppose you knew interest rates were about to get announced at 9am local, and knew if they went up, went down, or held, how you would need to be on a market position was known to you. You want an algorithm to get the news and then execute: buy, sell, or do nothing.

Advent of Alpha Day 23: Inferring a Book

Fri, 20 Dec 2024 19:11:17 +0100

Let’s suppose you have two selections in a market - say, draw no bet winner markets in football.

You notice that selection A has had £100,000 matched since the market opened and is currently 2.0/2.02 back/lay spread.

Selection B has had £90,000 matched since the market opened and is currently 1.9/1.91 back/lay spread.

That means for some reason a selection that has traded nearly 53% of the money is longer than the selection trading 47% of the money: that doesn’t make a lot of logical sense.

Advent of Alpha Day 22: BSP Auction Games

Fri, 20 Dec 2024 19:10:48 +0100

What do you know about BSP auctions?

I think there are two things you should know.

The first is that your bet directly affects BSP. This seems obvious but if you’re betting small you might not think much about it.

There’s a lot of alphas in looking at how your bets do or could influence those markets with low amounts going into auction (I’m thinking NZ horse racing, but to be honest UK premiership football is not great). £10 liabilities can move prices in a way that leads to a book being overbroke (so if you think a bot is around doing that you can exploit that).

Advent of Alpha Day 21: Getting Under Betting Limits

Fri, 20 Dec 2024 19:10:27 +0100

What’s the minimum stake you can lay or back on the exchange in the UK?

The web interface insists on a backer’s stake of £1. It used to be £2. You can do better.

An SP bet has to have a lay liability of £10, meaning that at laying at odds of 1000 you can have a backer’s stake of just 1p.

On the API you can bet to a target size (not supported by flumine), of £10 liability or payout regardless of bet size: the minimum is £10 target, you choose how the API assesses your size.

Advent of Alpha Day 20: Best Price Execution Fun

Fri, 20 Dec 2024 19:09:02 +0100

When you take a price ticks away from the best price available, you will get matched at that best price by default.

But you can turn it off at an account level in some jurisdictions. Why should you ever not want best price execution?

Sometimes you will be betting and the price will move between you sending the bet and it getting matched, and you will then want to cancel your bet (think in-play and betting on an outcome which becomes less/more likely during the bet delay).

Advent of Alpha Day 19: Bayesian Thinking One Offs

Fri, 20 Dec 2024 19:08:47 +0100

I’m not a statistician by training, I got into this by problem solving through code. However, when I first read about Bayesian statistics, it made me excited in an intuitive way that frequentist statistics just didn’t.

Most people learn frequentist stats at school: given a set of observations, figure out some properties that tell you about the shape of those observations (mean, mode, median, variance, and so on), and use those numbers to help make some sort of analytical model that can help you predict ranges in which future observations should fall.

Advent of Alpha Day 18: Gait, Weight & Computer Vision

Fri, 20 Dec 2024 19:08:31 +0100

ITV Racing love a paddock judge. I think I get why: one of the best reasons to actually go racing is to see lines of muscle conformation, how a head is carried, where the feet are being placed, etc., so by offering that via an expert is helpful.

It also feels a bit mysterious: how many people know how to look at a group of horses and tell you based on how they’re walking around a paddock, how well they’ll race each other? Not many. I’m not sure even the people who ITV employ, if we’re honest.

Advent of Alpha Day 17: Form Calibration to Markets

Fri, 20 Dec 2024 19:08:14 +0100

Newspapers that want to give the impression of data density like to print a form line, with the most recent result at the right, going back to the left. On race cards “-” indicates a break to the last season and “/” indicates a break to previous seasons before that.

On a race card you might see:

U4/301-211

This indicates the horse won the last two times out, came second before that and last season had a 3rd place before not coming in the top 9 and then on their last race of the season finishes the season with a win. The previous seasons had an unseated rider before a 4th place.

Advent of Alpha Day 16: Games Trainers & Jockeys Like to Play

Fri, 20 Dec 2024 19:07:55 +0100

This is the last of the “Games X like to play”, articles. It should tie together some ideas from previous days.

On day 13 I talked about the fact the participant mindset - particularly with animals - is under-valued. On day 14 I talked about training facilities often not being considered, despite them directly contributing to event preparedness and suitability. On day 15 I talked about how patrons can cause daft decisions because of their patronage. But the last group also have opaque interests too: trainers, managers, coaches, and in the case of racing, jockeys.

Advent of Alpha Day 15: Games Owners Like to Play

Fri, 20 Dec 2024 19:07:34 +0100

I was in the Betfair hospitality tent on the Tuesday of Cheltenham 2017, enjoying a drink and listening to their invited guests - Paul Nicholls and an assistant to an Irish trainer who I won’t name, to protect the guilty.

An invited guest asked a question about a horse within the stables of the Irish trainer and asked if the price was “a bit too generous”.

“Good grief, no. That horse should not be within 100 miles of Cheltenham. We’ve brought him because the owners want him here, and they’ve paid for that privilege to have a nice day out with an owner’s badge”.

Advent of Alpha Day 14: Games Horses Like to Play at Home

Fri, 20 Dec 2024 19:07:15 +0100

Day 13 was about the games horses like to play on race courses: racing prominently and staying ahead, or slowly pipping off other horses, with the jockey helping time the run so as to be ahead as they cross the line.

But what about at home in the stables? Or, more specifically, their home training regimen?

Do you know what training facilities the top 5 jump and flat trainers have at home? If you don’t, are you sure you understand what they’re sending to those events you are betting on?

Advent of Alpha Day 13: Games Horses Like to Play

Fri, 20 Dec 2024 19:06:50 +0100

How much time do you spend thinking about how a sporting event is going to play out before it starts? Perhaps if you’re thinking about first goal times or scorers, you might think about early domination. But do you then think about reactions to those goals?

It seems obvious in hindsight how each action caused a reaction, which then caused a further reaction. The drama of sport is in the sequence of to and fro, back and forth. The fact it isn’t scripted makes it all the more interesting - it is why sport is interesting.

Advent of Alpha Day 12: Event Physics

Fri, 20 Dec 2024 19:06:32 +0100

For the next few days we’re going to get into the weeds of a sport I love, and have loved for a long time. If you’re not into horse racing, you might want to walk past, but I suspect there are parallels in other sports. In today’s article I’ll actually discuss such an example.

What is happening when a horse runs around Cheltenham?

I honestly think people abstract it all far too much. They look at the form, they look at how the horse has run on particular ground, or after X days away from course in the past if they can, and then try and translate it to today’s event and figure out whether it is well suited or not.

Advent of Alpha Day 11: Calibration

Fri, 20 Dec 2024 19:06:18 +0100

Let’s suppose you think the price of something should be 2.0 - how do you test if that method of deriving the price is any good?

Calibration.

There’s a few methods for this, but I think there is alpha in just looking at the outputs from the simplest method, particularly when looking at how good somebody else’s prices are.

Take 1000 events, take the prices and turn them into implied probabilities (2.0 = 50%, 5.0 = 20%, 1.5 = 66.6%, and so on). Now look at how often that outcome occurred, and count it as a percentage. Plot it as a line.

Advent of Alpha Day 10: The Trend is your Friend

Fri, 20 Dec 2024 19:06:07 +0100

If you’ve ever read a book on day trading or technical analysis, you’ll be familiar with the idea of following trends.

Of course, finding the trends, and figuring out how much to put into it is the hard bit. In the context of a betting market which behaves more like an options market with a fixed time of expiry in the future, you might wander if trends are related to fundamentals - and they often might (a selection moving down through 1.01 as the event nears its finish may or may not be a good trend to follow).

Advent of Alpha Day 9: Delta T

Fri, 20 Dec 2024 19:05:53 +0100

A few days back I mentioned William Poundstone’s Fortune’s Formula as a great narrative non-fiction telling of Kelly criterion’s creation.

Another book Poundstone wrote that I’ve enjoyed is How to Predict Everything (this was its title in the UK, I believe it has another title in other markets). It is on the surface an introduction to how Bayesian probability can help you predict anything (we’ll return to this another day, perhaps).

Advent of Alpha Day 8 Markov Chains

Fri, 20 Dec 2024 19:05:39 +0100

Markov Chains describe a sequence of possible events based on the probability of moving into a given state based on your current state.

They are the foundation of a great deal of thinking in the reinforcement learning World, but you don’t need all that to start making use of them.

Suppose a selection is currently priced at 1.5/1.51.

What is the chance of it moving to 1.51/1.52? Or 1.6/1.61? Or 1.8/1.81?

Advent of Alpha Day 7: Hedging

Fri, 20 Dec 2024 19:05:24 +0100

You back a selection at 2.0, your bet is matched and now it’s sliding in. 1.95… 1.9… 1.8… 1.7… should you lay it off?

There’s a lot of discussion about this in some circles. One group is “never green” - if you had edge at 2.0, ask what’s changed now? Giving money back to the market means you reduce your profits.

I’ll say you should often consider hedging. Often enough in fact that on many strategies I’ll have a ’limit on close’ order or ‘market on close’ order to close my positions. Why?

Advent of Alpha Day 6: Kelly

Fri, 20 Dec 2024 19:05:12 +0100

John Larry Kelly Jr. probably created the most important equation related to the industrialisation of Information Theory, probability and wealth management in history.

Don’t believe me? Well, walk around Wall Street or the square mile, look around, and realise until Black, Scholes, Merton, the most important formula doing the rounds to build that place was Kelly’s 1956 work in showing how to optimally size a bet.

The simplest version of explaining Kelly is that the proportion of your bank as a percentage is your edge, expressed as a percentage, divided by your fractional odds. Got 2% edge at evens? Bet 2% of your bank. Got 0.5% edge at 7/2? Bet 0.5/3.5%, or 0.14% of your bank.

Advent of Alpha Day 5: Latency Arbitrage

Fri, 20 Dec 2024 19:04:55 +0100

“Latency Arbitraged” is a fancy name for court-siding.

Court-siding got its name because the first sport where this got a lot of traction was tennis, where young fools were paid to turn up to events all over the World with a mobile phone and to hit the buttons as a point was scored, many, many seconds before the official score was updated - enough to beat in-play delays on exchanges. I’ve read books by people who did this that made me think it’s a hard, dangerous and horrible way to make a living (tennis authorities do not like it, and local law enforcement have been alleged to get creative about stamping it out).

Advent of Alpha Day 4: Derivative Arbitrage

Fri, 20 Dec 2024 19:04:43 +0100

Yes, this idea of daily posts hasn’t worked quite as well as I had hoped. A mixture of illness, travel, work and other stuff meant I’m rather behind. But let’s crack on.

Due to time pressures, these articles are going to get a little more “bite-sized”, but if something grabs you and you want to discuss it, there will be threads on the betcode slack.

“Derivative Arbitrage” is not a technical term I’ve ever heard anybody ever use, I just don’t know a better one when trying to describe it in a two word phrase. The core idea here is that prices in one market can help you price another.

Advent of Alpha Day 3: Basic Arbitrage

Mon, 02 Dec 2024 19:57:48 +0000

Arbitrage (or “arbing”), in a basic form the simultaneous (or near simulataneous) buying and selling of the same or equivalent assets in different markets in such a way as to guarantee a profit. Buy low, sell high, just like market making, but you’re doing it in two places at the exact same time.

If a friend is offering to sell you his old games console for £100, and you know somebody who has been looking for that exact same console with a budget of £150 in mind, you have the ability to arb it, lock in £50 and make both your friends happy. This is boring stuff, but it’s simple and useful.

Advent of Alpha Idea 2: Market Making

Mon, 02 Dec 2024 10:45:53 +0000

Market making and the idea from Day 1, Bookmaking, are synonyms in the eyes of some in the context of a sports exchange. Traditional market making on a stock or commodity is buying at one price (high) and selling at another price (low).

When you “buy” on a sports exchange you’re “selling” everything else (assuming one winner on the market), so bookmaking by laying (selling) one selection is in effect buying everything else, because of how the market is settled.

Advent of Alpha Idea 1: Bookmaking

Sun, 01 Dec 2024 19:05:11 +0000

Welcome to the first day of the Advent of Alpha.

I know a lot of people will see the title and think “Really? Not exactly innovative, is it?”. Using exchanges to make books might be “obvious”, but it’s devilishly hard, and the ideas you have to start thinking through are good foundations for other ideas.

The moment anybody who knows something about betting sees an exchange there are two thoughts that go through their mind: “these odds look a lot better than the ones I get to back at down the high street bookmaker”; and, “Wait, you can lay? You can become a bookmaker?”.

Welcome to The Great Triviality, and the 'Advent of Alpha'

Sun, 01 Dec 2024 18:59:32 +0000

Welcome to The Great Triviality (named for a quote from the late, great Phil Bull, founder of Timeform).

This blog is dedicated to exploring algorithmic trading strategies on betting exchanges like Betfair. From now on I’ll refer to this activity through the over-used word “trading”.

There is a community of people - some doing this professionally - who have gathered on the betcode moderately active Slack channel, to discuss all things trading. The betcode group has as an obvious focus on the superb open source Python library flumine. Other libraries exist, but this one seems to have gathered the largest community discussing trading, even if by most accounts many people are actually writing other tools to trade.