Is Turdle Racing Worth It? A Statistical Analysis
Turdle Racing (TR) in Meridell, the considerably smellier cousin of Poogle Racing, seems to be often overlooked - probably because unlike its Faerieland counterpart, it doesn't have an avatar to its name. However, its lack of glamour doesn’t mean it isn’t worthy of recognition! I myself came across it while investigating older parts of the sites, in an attempt to find new dailies or games that could be profitable. Many of the older games on the website were designed to bring Neopoints into the economy, when the site was just starting up (see also: Stock market, Food Club). I theorise that since TR is a very old game that has not been updated, it is possible to play in such a way that you can consistently make a profit by playing it. This investigation aimed to determine whether or not this is true by analysing the results of 441 Turdle races which I played over about 5 months.
Please note that, in this article, “profit” generally refers to “profit on average” or “profit in the long run”. This is a luck based game, so it isn’t about what might happen for each individual game. Rather, we are theorising about general trends.
How to play
Turdle Racing is a luck and chance game in which you bet on one of five Turdle Petpets in a race (which I will call T1, T2, T3, T4, and T5). First, you select which of the five Turdles you want to win and enter your bet amount. You can also choose to feed the Turdle a food item from your inventory. Then you watch the race to see who wins. If your Turdle wins, you win back the amount you bet, plus your bets multiplied by the listed odds of that Turdle winning. You can bet a maximum of 1500NP per game, and play a maximum of three times per day.
Example: Say you bet 1000NP on Turdle 3 (Smelly) and it wins. You win back your original 1000NP plus your bet amount multiplied by T3’s odds (5:1). So the total amount you win would be 1000 + (1000x5) = 6000NP. (If T3 lost, you would lose the 1000NP you bet)
The odds for each Turdle listed on the main betting page are as follows:
Table 1: Turdles and their corresponding listed odds
I initially decided to play and record the results of choosing the first three Turdles 3 times a day for 40 days, for a total of 120 trials per Turdle. Once I had finished those 360 trials, it was clear that the most profitable Turdle was probably T1, so I continued to play T1 3 times daily and recorded my results so that I would have more data and therefore more reliable results. In total, I played T1 201 times, T2 120, and T3 120 times. For each trial, I bet the maximum 1500N, did not feed, and recorded whether my Turdle won or lost.
Table 2: Frequency and profits of Turdle Racing wins
These results indicate that betting on Turdle 1 and Turdle 2 was profitable, whereas Turdle 3 was a small loss. Turdle 1 was most likely to win with a win rate of approximately 37%. Turdle 1 was also profitable with an average net gain of about 670NP per game - that’s about 2k profit per day if you were to place 3 bets on T1 daily. T2 won 25% of the time, which translates to about 375NP per game. T3 won about 16% of the time, which corresponds to a small loss of 75NP per game on average.
Explaining the results
In short, it appears that you make a profit from TR because the odds are displayed in a way that is unique to Neopets, ensuring that you win a profit on average over time.
Usually, a ratio in the form “x:y” usually means “x number of losses for every y number of wins”. For example, a likelihood ratio of 3:1 means you would expect 3 losses for every 1 win, ie a ¼ chance of winning. According to their listed odds, T1 (3:1) and T2 (4:1) had expected win rates of ¼ and ⅕ respectively. However, the experimental win rates were closer to ⅓ for T1 and ¼ for T2. This suggests that in Turdle Racing, listed odds of “x:1” corresponds to a “1 in x” true chance of winning. This is unusual to see in real life, but it does have precedent on Neopets; Food Club odds are presented in this way, where odds of 2:1 approximately means “1 in 2 chance of winning”.
You can demonstrate this by doing more rigorous statistical analysis* (specifically Binomial Tests). For example, this shows you that T1’s true win rate is very unlikely to be ¼ (which its 3:1 listed ratio would indicate) and very likely be ⅓.
(*I have omitted the detailed analysis from this write up because it took too long for me to simplify enough for an NT article. Sorry!)
When you win in Turdle Racing, not only do you win your bet multiplied by the odds but you ALSO win back your original bet amount. For example, if you win a bet on Turdle 1 who has listed odds of 3:1, you actually win *four* times your money back. If our results are true, and Turdle 1 has a true win rate of 1 in 3, this means betting on T1 wins you four times your money back one third of the time. This theoretically translates to 33% net profit over time. (That’s 500np per game if you bet the full 1500NP per game, or a maximum of average of 1.5K a day since you can play three times). It also means that hypothetically, all the Turdles are profitable: T2 would have a profit rate of 25%, T3 20%, T4 14%, and T5 11%. As you can see, Turdle 1 would be the most profitable, with a theoretical profit of 500np per game.
In our results, T1 and T2 reflected this, but Turdle 3 did not - in fact, T3 showed a small loss, and was quite close to the originally hypothesised win rate of ⅙. However, this could be explained by the small sample size and low probability of T3 winning; in 120 trials, the number of wins is small enough that a single win can skew the data a lot.
This study was prospective - I planned the method and hypotheses first, and then tested them. This reduces the chance of bias compared to retroactively analysing results that were collected for some other purpose. All the results were collected by me, and therefore (hopefully) consistent and accurate. Assuming that Turdle Racing is the same for everyone (and it almost definitely is otherwise that would be weird) this study has good internal and external validity; that is to say that the results are consistent, reliable, and applicable to other people. The sample sizes for T1 and T2 were reasonably large to get meaningful data.
Even with 120 trials, this is a fairly small sample size especially for Turdle 3, which has a small chance of winning.
Due to the time it takes to collect results, I only focused on the first three Turdles. A more comprehensive study could test out T4 and T5 also. With just three data points, you can’t really verify any trend.
Areas of further research
If you wanted to learn more about Turdle Racing, here are some questions to investigate
Does feeding the Turdle improve your chances of winning? If so, do particular types or rarities of food help more than others?
Are Turdles 4 and 5 profitable?
Can these results be improved on by having larger sample sizes?
Please let me know if you do test any of these out - I’d love to see your results!
According to our results, Turdles 1 and 2 are profitable to play, whereas Turdle 3 is not. In this study, Turdle 1 was the most profitable, netting about 670NP per game if you bet the maximum amount of 1500NP per game. Turdle 1 was also the most likely to win, winning about 37% of the bets I placed on it. I would recommend placing 3 bets of 1500NP on Turdle 1 as a daily, as this study indicates you will profit quite consistently over time.