Looking for a perfect Tinder Date? This bit of math might help
[Update: The strategy may not work anymore because Tinder has changed their algorithms significantly and has added some new features since. But keep reading anyway, it’s kinda funny]
Well, maybe it isn’t that deep, statistics shows a lot of Tinder users are not there in search of a potential mate. Either way, I would like to present a report of an experiment I did using a Tinder simulator I created using Python.
If you are a computer science or math enthusiast you might want to keep reading from the beginning but if you are here for just the strategy and don’t want to read how I devised it, I suggest you skip to the end of this story.
The 1/e Stopping Rule
In mathematics, the theory of optimal stopping is concerned with the problem of choosing a time to take a particular action, in order to maximize an expected reward or minimize an expected cost. Apparently these problems are very common in different areas of statistics, economics and mathematical finance. The Secretary Problem is one of the classic examples of optimal stopping problems.
The basic form of the problem is the following: “imagine an administrator who wants to hire the best secretary out of rankable applicants for a position. The applicants are interviewed one by one in random order. A decision about each particular applicant is to be made immediately after the interview. Once rejected, an applicant cannot be recalled. During the interview, the administrator can rank the applicant among all applicants interviewed so far, but is unaware of the quality of yet unseen applicants. The question is about the optimal strategy (stopping rule) to maximize the probability of selecting the best applicant.” [wikipedia]
The difficulty is that the decision must be made immediately and can’t be deferred to the end.
This problem was resolved by some mathematicians long ago. An easy analysis implies that the optimal win probability is always at least 1/e (around 37%).
Which means, the optimal strategy to choose the secretary from a group of candidates is to interview the first 37% and reject all of them regardless, then interview the rest until you find the one who is better than all the 37% previously interviewed.
How is this significant in our case? If our goal was just to devise a hack to super like the person who you would like the most, then going with this strategy would be the best possible way. But actually, the way Tinder works is much more complicated. While the idea might sounds flattering in theory, being Super Liked can be very creepy in reality. Anyway, statistically it increases your chances of getting a match by x3 (According to Tinder). It should also be considered that we are not dealing with an accept/reject problem, because there are two different ways you can like a person on Tinder.
Considering the new constraints, intuitively we may conclude that the stopping point in our scenario will be less than 37% (There is no need for a proof here since this assumption does not affect our future conclusions).
Our task is to find the magic number (the optimal stopping point) that will maximize a person’s chance of getting a “perfect tinder date”.
I created a computer simulation in an attempt to find that magic number. Sadly, I had to assume a lot of things because there was not enough reliable data available. One of these assumptions that I took from a random source was, Tinder users swipe right on 80% of the profiles they are exposed to (I fixed the threshold based on this, on uniformly distributed attractiveness scores which probably should have been normally distributed but fortunately, this whole assumption was found not alter the result in the end anyway).
I also assumed the probability for being liked back is the same regardless of the attractiveness score of the person you like (That was a wild assumption but is not likely to affect our result. If you assume a inverse correlation between the probability and attractiveness score — that’s the only other logical assumption as long as we don’t have reliable statistics — it will end up averaging out because the objective is to increase a weighted sum of probabilities where the weight is the attractiveness). The constraints included the swiping limits.
But the first question I tried to answer was if it would be advisable to swipe left on profiles to find the one who you would like the most, given that you have a super like left but your have exhausted your swipe-rights. The answer might be yes, because the number of profiles on Tinder is practically inexhaustible. So I decided not to simulate that scenario. Finding the magic number was the original task and I just kept at it.
But how would we find it? My way of approach was to create a sample space of people with corresponding values (randomly given) of attractiveness scores, run the simulation with numbers from 0 to 100 as our magical number and analyze the respective results (sum of weighted probabilities).
What I found after iterating the simulation 10,000 times was that the avarage (of 1,000 simulations) stopping point was consistently (9 out of 10 times) observed to be 30!
In conclusion, for best results, you have to super-like the first profile you find more attractive than all of the first 30% profiles you visited! (Who keeps track of that lol)
(It was a very naively designed simulation. Don’t even trust me!)
Tinder (free) allowed you to swipe right 100 times (in 12hrs) and super like once (a day), but now the numbers appear to be lower than that for most users. It is observed to vary from user-to-user so you have to change your strategy accordingly. The following numbers are based on the 100 swipes one.
- Don’t super like any of the first 30 (30%) people Tinder shows you.
- After going through (regardless of the number of people you swiped right or left) 30 profiles, super like the first one who you find more attractive than any of the 30.
- Exhaust your super likes!, no matter how many profiles you have to swipe left to find the right one.
These tips might help you optimize your chances, but ultimately, the whole thing depends on who you are and how you present yourself.