Cardiffenio: analysis of his latest results

Cardiffenio: PRO Tipster

The tipster exposes his vision about his recent streak

I never had to justify my way of betting before, and that was one of the things I liked the most about being a tipster: that freedom to earn money and being able to make money for others without having to give explanations, just based on the tips and numbers. However, my recent change of platform coupled with the biggest drawdown since I predict forces me to justify my numbers in some way.

First of all, I want to make clear that my betting system has remained constant since I started to forecast in October 2014, following a statistical method (corrected with news and assumptions on lineups etc) that I have been using since 2013 with few changes.

The biggest change I experienced was a reduction of stakes in bets since my last website made me PRO Tipster. The average stake went down from 3.5 to 1.75 progressively, unconciously being too cautious once I realized there was people that was following my tips with their own money. Once my customer base grew I had to give up some leagues of low liquidity due to the difficulty to follow them for the followers (Paraguay league, Peru league and Segunda B, mainly).

If we stick to the leagues in which I can send send tips today, and using flat stakes to avoid any suspicions, this would be my record since 2014:

You can see (flat stakes) a Drawdown of 25.32 units in the last weeks, and a yield of 3.9%.

According to my way of selecting stakes, for me a stake 1 is a bet with a theoretical yield between 3 and 5%, stake 2 would be between 6 and 8% and stake 3 above 9%. The historical distribution of these stakes is not as expected:

Stk1: yield -4.8%, 672 bets



Stk2: yield 7.4%, 797 bets



Stk3: yield 7.8%, 790 bets.


According to my estimations, the results of stake 1 are negative, much lower than expected (I hope that in the long term they will have a yield of between 3 and 5%). The stake 3 are below expectations, and stake 2 is higher than expected. In any case, due to the number of bets, it is still early to think that these numbers are far from definitive.

If we stick to the pvalue, you need way too many bets to rely on a tipster. In my case, being close to 3000 picks could be almost enough to accept some advantage over the market, but those long-term numbers can´t do anything to analyze the streak of recent months. For me the best way to have a qualitative analysis is to compare the bets of a tipster against the market consensus, is take the reference of closing prices in Asia and see if the tipster in question usually recommends prices above or below those prices.

Using the closing price of Pinnacle Sports (or another reference page) we can estimate the theoretical gain of a tipster in a given market.

There may be many criticisms regarding this methodology, for the following reasons:

-The market is not always efficient, when moving according to the money invested by each person, and not how much each person knows.

-Efficiency is not the same for different leagues.

-Deviation of the price due to bets of the general public (bets to favorites, etc.)

However, I believe there are enough demonstrations about closing price efficiency to disregard those small inefficiencies and use this qualitatively.

If we focus on the tips that I have sent since the change to Inbetsment, we can see the following chart, with the expected gain associated:

(note: the expected gain is calculated by multiplying the stake and the difference between tipster odds and the actual price that Pinnacle calculates. You can calculate this last price dividing the final price by 0.98)

Example if we bet 2 units to a DNB to 1.95 and Pinnacle closes in 1.75 the theoretical gain would be 2 * (1.95-1.75 / 0.98) = 0.32ud] ):


The blue line indicates the units lost, the red line the loss units if all the bets had been 1 unit, and the yellow line represents the theoretical gain based on this market consensus. That theoretical gain was 3.1% in these first months of the year, below my usual gain but not so far from what I have been doing in previous years.

Focusing on all stake 2 bets you can see how a lot of bets have been lost even though the theoretical gain was quite good:


Although I do not have the full data for all picks since the beginning, the theoretical gain in my bets has been relatively consistent around 5% in these leagues, which has generated a yield of around 6.5% in more than 5000 personal bets. It is a good reference to follow the theoretical gain because it tells you how the variance is influencing your results.

It is also interesting to note that, although the numbers might seem to indicate otherwise, I think my stake 1 bets are good enough to keep sending them, and that in the long term they should show returns. This can be visualized with expected gain versus closing prices as well:



So, in conclusion, I think I can say that despite the bad numbers of the last few months, there is still reason to believe that my bets retain some value and that in the future they should see positive results. The market will become more efficient and the yield should go down, but not in such a dramatic way.

Simulating the results obtained in the long term, which is to assume an expected yield of -5% for Stake1, 7% for Stake 2 and 7.8% for Stake 3, a poor streak such as that which has occurred in recent months to 170 has a 7.5% probability of happening. If we think in a fixed value of all bets of 6%, the probability of a drawdown of more than 40 units lost is 4.7%, that is, something that will happen from time to time in a history of more than 2500 picks.

In any case, that doesnt means that the drawdown is finished. As in the "gamblers fallacy", the drawdown has no memory and there is still a probability that the drawdown will continue. That is, even assuming a market advantage of 6% constant, there is a probability of 4.7% of there is another drawdown of more than 40 units lost in the next 170 bets. But let's hope it finishes soon and more normal returns come back.