『當家精神』爆冷有跡可尋?
(Please scroll down for the English version)
12月20日的沙田賽事，「當家精神」初出即以47倍爆冷取勝。除了幕後知道這匹馬被嚴重低估外，我們可以看看賽前大眾對牠有什麼看法：
賽前最後一場試閘輸7個馬位。
各大馬經專家「貼士」
於賽前google「當家精神」，唯一對牠有正評的是東方日報中有一句-4試大閘而臨必有計。
從以上資料可得知，沒有馬後砲下大眾對牠的唯一正評是試得多閘。
我們看看如何用Python分析一下新馬的勝率吧。
流程
Step 1 — 在數據庫中filter走其它馬和新馬賽（GRIFFIN），然後看看有多少匹新馬。另可看出大部份馬都是52/57分評分。不清楚原因的話可以看這。
Step 2 — 和之前的文章一樣，我們看看大眾對新馬的勝率計算得怎樣。
左圖熱門區域的WIN比PUBLIC_PROB大，可看出大眾有點低估熱門的新馬。
Step 3 — 活用pandas + numpy的 where 做一個簡單的betting strategy.
過去十年，如果只下注10元給20倍以下的馬匹，最後會輸了513元 （-4%回報）。不計算馬會16.5%的抽水，其實這麼簡單的策略就比Random Bet（-16.5%回報）高出不少，可見賽馬的彩池還是不太efficient
最後，有興趣的話可以看看PPG/PP新馬的數據有什麼玄機、試閘多少次真的對勝率有影響。有機會再講。
A new horse -Carroll Street with winning odds 47 crashed out other competitors in the horse racing competition conducted on 20th December in Shatin. Insiders would know the horse is being undervalued, but let’s see what’s the public view on this horse before the race.
7Ls off pace in the last barrier trial before the competition
Tips from various horse racing newspaper
The only positive comments on Carroll Street before the race is “Possible strategy with 4 barrier trials“ from Oriental Daily
From the above information, the only positive comment on this horse is numerous barrier trial.
Let’s see how we can do an analysis on the new horses winning odds via Python.
STEPS
Step 1 — Filter out other horses and Griffin in the database, and let’s see how many new horses are there in the database. Other than that, you may also notice most of the initial ratings for the new horses are 52 or 57, you may find out the reason in this link.
Step 2 — Similar to preview articles, let’s see what’s the public place odds on new horses.
From the left image, the smaller the winning odds, the actual winning probability is higher than the public probability, indicate the public undervalued some hot new horses
Step 3 — Make good use of Pandas + Numpy‘s where to build a simple betting strategy.
For the past 10 years, if we only bet HKD 10 for horses with winning odds lower than 20 times, you will be losing HKD 513, i.e. -4% return. Taking into consideration of 16.5% commission from HKJC, this simple strategy generates a much higher return than random bet, i.e. -16.5%. This proves the betting pools are quite inefficient.
If you are interested, you may further study if you can draw any insights from PPG (Privately Purchased Griffins)/ PP (Privately Purchased) new horse data or what’s the correlation between the number of a barrier trial and winning odds. Stay tuned.
