Deeper down the Morning Line rabbit hole we go

Last time out we took a look at Morning Line (ML) prices and what they might be telling us in the game of PhotoFinish.Live. I am sure you read and understood everything discussed so no need to recap here. You can give Google Analytics something to do by reading the first ML article here.

Okay, now we are all up to speed, why are we back debating ML again? Well further analysis has posed more questions than answers, and rather than just give answers, let's follow Alice down the rabbit hole (See the image above is the Red Queen from Alice in Wonderland, that's where "down the rabbit hole" comes from. I spent more time on that than writing the rest of the article!!)

Last time (I know I know) we figured out that ML prices are 50%+ dictated by attributes (which we can't see until a horse is retired) and preference stars with weightings varying by distance. We also found out that ML prices are fairly accurate, and win in roughly the proportion they should implied by the price. Favourites don't always win and if the favourite is a 4/1 shot (5 in decimal odds) it will win 20% of the time. Ish. All this analysis is at a population level - ie across all horses. I have had a stimulating week "debating" with various people on the servers of discord about magical horses who do not conform with my broad brush theory. So let us take a look at individual horses, and see what is going at the micro rather than macro level.

This analysis was conducted on data that was updated on 11th August 2024 just in case you are reading this in 2029 when Crown is invariably at $10.

Underdogs or Overrated

Let's divide horses into a 2x2 matrix of wins versus being favourite. If this was BCG (Boston Consulting Group) they would have charged you $1MM for this analysis.

Obviously real life is messier than a 2x2 matrix, but we are not really interested in every horse - we want to look at the extremes. Why are some horses always the ML favourite, yet hardly ever win? Why do some horses frequently win yet are never ML favourites?

The actual plot is below, and like any cloud of such dots has a low correlation coefficient (of 0.15). So there is no strong relationship between winning and being the ML favourite. This should not surprise us. Most favourites are 6 to 8 in decimal odds and as such will only win 15-20% of the time. But we are in Simpson's Paradox territory now (good name for a horse) and down the rabbit hole we must go.

Let's start with the good horses. Pop quiz. Which horse raced 66 times and was the ML favourite in 65 of those races, and 2nd favourite in the other one?

It was money printing machine, Mr Wonderful. An S+ grade horse with +3 subgrades above S+, which is extremely rate, coupled with SS in speed and stamina, and 7.5 preference stars.

This list of good horses contains many of the perceived champion horses in the game, with a mix of newer classic winners and older ones. Many of these horses were the favourite in 95%+ of their runs. The game was telling you they were good horses, and the majority of the time they won. But you can't rely on just the ML prices because of the presence of Underdogs.

What makes Underdogs win?

Underdogs are horses that win frequently but are rarely ML favourites.

Just today, Alva, 21 runs, 9 wins and only twice favourite, won a grade 1 qualifier despite not being fancied by the ML prices as shown. This is a typical scenario for an Underdog:

Why is Alva constantly underrated? We can't see the attributes yet, so we can't be certain of what is driving this, but there are horses that win regularly despite not being the favourite. Hold that thought for a minute whilst we look at the Overrated.

Which horses are Overrated?

Horses that are often ML favourites yet rarely win at Overrated. We've all been guilty of that false confidence when we look at race our horse is entered in and see it's the ML favourite and think we have it in the bag. Why does ML rate them so highly and why don't they win? Let's eyeball some of them.

Lady of the Wood has good attributes, 5.5 preference stars, and is a +1 subgrade S+ horse of the LDF archetype. With SS- for start and speed and S+ for stamina and finish, everything looks good. But why does it rarely win? It was run at the appropriate distances. Let's take a look at the bloodline. What is striking is the consistently low rarity scores in its bloodline.

Three 1/1s and a number of other sub 100 nfts. If we used this as a proxy for bloodline quality, it would be a fairly good score. If you look at some of the others on the list, you find similar patterns. Below is Sully, winless in 29 races but 22 times the favourite. What do you notice? 2 out of 4 1/1s and a low green would give this horse a low weighted score on rarity too.

It would appear that horses with high quality bloodlines are highly regarded by the ML prices, but that doesn't always translate into a horse with winning attributes. Which brings us back to Alva the Underdog. Alva has three great grandparents without 1/1s (but does have matching green background semi-rares). Clearly the algorithm doesn't rate her chances as high as it should because it possibly thinks the bloodline isn't as strong. This is speculation on my part, and veering away from the usual statistical analysis. More analysis is required. You can use the lists of horses shared here to conduct your own analysis. Maybe you have a better insight into some of the horses featured and can provide me with more background. Or maybe you have found an outlier that proves a pattern. This is the beginning of ML analysis, unfortunately, not the end.

What conclusions can we draw?

Last time out we figured out that 50%+ of the variance in ML price can be explained by attributes and preference stars using retired horses and some regression analysis. I think that bloodline is one of the other elements that would explain more of the variance. We just need to build something to quantify the quality of a bloodline. Maybe throw matches in too. That will take some work.

Horses don't win because they are ML favourite - the vast majority of the time the favourite loses - just like in real life. However, some horses win often, because they have superior attributes and preferences, with strong bloodlines. It's the fact that a horse wins often that predicts the future far better than ML prices. Yes, ML prices are very useful if you are handicapping horses yourself, but past behaviour is the best predictor of future behaviour, and if a horse is winning frequently it will probably win again in the future.

There is obviously a lot of noise in this data - just see the scatterplot. Some horses were run to farm crown stimmies. Not all archetypes were created equal. Some owners are better at placing their horses in races they can actually win. Others take a long time to find the ideal conditions and optimal distance.

Analysing ML prices is a rabbit hole, every time you think you have an answer, five more questions arise.

Join the fun and put these insights into practice at PhotoFinish.Live and if you are considering starting your own stable please consider using my referral code: PADDOCK or just click on this link: https://signup.photofinish.live/?referralCode=PADDOCK

Please remember this is a web3 game where your spend your own money. Nothing I write about should be considered financial or investment advice.

Other blog posts:

• What can ML prices tell us: https://aipaddock.com/skill-luck-and-morning-lines

• Finding the optimal distance: https://aipaddock.com/understanding-optimal-distances

• Understanding breeding: https://aipaddock.com/understanding-the-most-successful-breeding-horses-in-pfl

• What are subgrades: https://aipaddock.com/what-are-subgrades-and-how-do-they-work

• Fastest horses: https://aipaddock.com/who-are-the-fastest-horses-in-the-game

• Racing profitability: https://aipaddock.com/how-much-profit-do-you-make-racing-horses

• Trueskill pvp ratings: https://aipaddock.com/which-horse-is-the-goat-in-pfl

• Breeding: https://aipaddock.com/trying-to-understand-breeding-in-pfl

• Evaluating horses: https://aipaddock.com/how-good-is-my-horse

• Understanding performance: https://aipaddock.com/understanding-the-true-performance-of-your-horse

• Do stars matter: https://aipaddock.com/how-much-do-preferences-matter

• FF Rating vs Finish Time: https://aipaddock.com/the-difference-between-ff-rating-and-finishing-time

• Are horses getting faster: https://aipaddock.com/are-pfl-horses-getting-faster