Performance Marketing vs Online Poker - Variance and Standard Deviation (6)
Performance Marketing vs Online Poker - Variance and Standard Deviation
For the introduction to the topic of AM vs Online Poker, please refer to this thread 
LAYMAN'S DESCRIPTION OF VARIANCE AND STANDARD DEVIATION IN POKER AND AFFILIATE MARKETING
One of the most important concepts to understand in affiliate marketing is variance. It just so happens that it's a vital concept for poker as well, and understanding it gives former poker players a leg up the average person trying out affiliate marketing. In this article I will try to translate some of the maths talk to more familiar terms and explain what this means for the results you see in your campaigns.
Let’s start off by defining variance, then we will break it down to what it means for our purposes.
In probability theory and statistics, variance is the expectation of the squared deviation of a random variable from its mean, and it informally measures how far a set of (random) numbers are spread out from their mean - taken from Wikipedia.
In layman terms, or intuitively, it’s a measurement of randomness. You have a certain average expectation for an event, but whenever the result deviates from the expectation, variance is involved. Luckily, we don’t need to calculate much for our purposes, but experienced poker players, after having played millions of hands online, usually develop a 6th sense for what is just variance and what is not.
STANDARD DEVIATION
Variance is also the square of the standard deviation. Standard deviation is used to quantify the amount of variation of a set of data values.
Say you are a poker player who on average wins 5bb/100, meaning 5 big blinds for every 100 hands played. This is the normal distribution of the results, with a typical standard deviation in No Limit Hold’em of 80bb/100. Translated, it means that although the average (and mean) of 5bb/100 is the most likely result, you will often find yourself deviating from those results. And this will happen purely because of random events.
What does this mean for affiliate marketing?
In affiliate marketing, most people talk in daily profit. There’s nothing wrong with doing that if you’d take into account variance, and talk in average daily profit, or your expected daily profit (coming from EV = expected value, very used term in poker). However, most people talk about their best days (or best 100 hands poker session), so they talk about the extremes when they find themselves at the far right side of this normal distribution graph from above.
One day for campaigns is like 100 hands of poker, maybe even less. In the grand scheme of things, you could win or lose 60 times more than your EV in that time period. Consider how much you deviate from your actual daily average profit if an advertiser decides to not pay you at all for a certain period of time.
If your expectation is already negative, that means you can have some seriously bad days. Even if your expectation is generally positive, you might still have plenty of losing days. This is true for both poker and affiliate marketing.
What makes poker players particularly strong here is understanding exactly that one day of results means nothing. And their future decisions are not affected by a data sample too small to be relevant to zero in on their actual EV/winrate/ROI.
That 6th sense for understanding variance is useful in all businesses, and many facets of life, but it has a particularly direct impact in affiliate marketing.
There’s something called The Central Limit Theorem (hat tip to John, the co-author of our poker book), which says that, and I quote John, “if you monkey around randomly often enough, you will always end up with a normal distribution”. In other words, if random events happen consistently, and often, they don’t matter.
In AM there are plenty of random events: zones get pulled, domains get flagged, offers get paused, rules change. Or, common favourable events: new traffic source, new zone, new offer. Although all of these have an impact on your immediate results, you cannot control them, and they are unpredictable. You can just call them random events. They also are sometimes unfavourable, sometimes favourable. This means your results will fit in a normal distribution, and have a standard deviation. Depending on your vertical, processes and work style, your average expectation changes, and so does the standard deviation. In translation, your actual results WILL vary much more compared to what your actual revenue/profit should be over a big enough sample that is less affected by these random events.
The better you understand these concepts and can apply them to your daily business, the more you can improve your mean, and better mentally handle all those swings. Not being results oriented but strategy oriented (or process oriented) is the main skill winning poker players have compared to the average losing player, and it may not be too different for affiliates who build a long term business compared to the "one hit wonders".
Part 3: Bankroll Management part of AM vs Online Poker
Part 4: Performance Marketing vs Online Poker - Masterminds and Study Groups
Open Q&A
One difference I forgot to mention is that perhaps affiliate marketing has more potential for so called "black swan events" (from The Black Swan, by Taleb), which means that highly improbable but with great impact is more common in AM. They would find themselves much further to the extremes of the bell curve.
Things like major regulation changes (like Google Play policies a few years ago), or a new vertical with lack of them appearing are very rare events but with great impact on ones expected value.
In online poker there was probably one very big such negative event, commonly known as Black Friday where big poker rooms went bust, and players lost millions of dollars.
Given that AM covers so many verticals, it's susceptible to such events in each and everything that connects the industry as a whole.
That doesn't make AM less profitable, it just underlines the importance of proper risk management.
Interesting series Manu.
I'd also like to mention the up to then unknown Chris Moneymaker who won the most important Poker tournament (Main Event WSOP) in 2003. He was one of the reasons for the Poker Boom afterwards.
In Poker, you can win millions with just a few $, so variance might even be higher than in AM, at least for MTT (Multi table tournaments) players.
Originally Posted by osmiumman
That guy said no to all deals, and just replied in chat "I want million". In the end he got it
Manu, great observations.
Although it's something that can be learned, I think a lot of successful affiliates are good at this intuitively. On the flip side one of the main frustrations I have with a lot of corporate performance marketers and consultants is that they often confuse what's variance and what's meaningful data i.e. they cut things too early or draw conclusions from data that doesn't seem "right".
Originally Posted by nemekn
Home > General > Affiliate Marketing Forum