Prop versus Algo – Like Minetta Tavern vs. Mickey Ds?

Boris Schlossberg

This month, after endless prodding from K I went back to prop trading. Surprisingly, it’s going well. Even more surprisingly I don’t hate it. I know, I know – for the past six months I’ve been screaming that the future of trading is rules-based algos, but trading prop has actually made me appreciate the art of the former as well really clarify the science of the latter.

Let’s agree on one thing. Prop is never going to go away. Human beings will always want to make decisions and furthermore the world of solely algos creates a recipe for disaster as machine choices either bring the markets to a halt (perfect equilibrium) where no trade can make money because all information is instantly factored into price) or worse markets crash with no support in sight – (2011 anyone?) but this time REALLY NO SUPPORT which could bring the whole system to it knees.

In any case, human judgment will always be a part of markets because markets are ultimately human enterprises. So it’s good to think about how prop differs from algo and more importantly how each approach has vastly different end goals that we need to understand.

This week the Beyond Meat IPO launched to a 150% pop on the first day of trading, so hamburgers are really on my mind. Besides who doesn’t love a burger? Last week in Cali I tried BM burger right next to a “real” one and I really liked it more than meat – but I will leave the sectarian battle to the food purists. What I want to propose today is that prop versus algo is very much like Minetta Tavern versus McDonalds.

Most of you probably haven’t heard about Minetta Tavern burger, but it is considered to be the best in the city. Again, I have no desire to fight a food war here (I am a guy who actually thought that burgers at Legal Seafood were some of the best I ever tasted, so I will leave my culinary judgment out of it), Instead I want you to think what it takes to prepare a perfect burger at a fancy restaurant. Each patty is carefully selected, checked a hundred times over for proper fat to meat content, individually cooked and dressed and presented to the patron in the most appetizing manner. Furthermore, if any flaw is discovered on the plate, the whole thing is quickly thrown away and a new one is made.

Prop trades are very similar. Each weekend I pour over all the charts, consider the event risk facing me, think about the current sentiment and positioning and try to put three perfectly curated trade ideas “on the plate”. Furthermore, if any of those ideas turn “sour” due to some fresh information in the market, I may decide to scrap the whole thing well before it hits my stop.

Now think about Mickey D’s. They are not trying to produce a great hamburger. They are trying to produce a “good enough” hamburger that they can serve a trillion times. That means that they take a lot of short cuts with their product, but on the other hand, produce the exact same patty in Dubai as in Detroit. No one at Mickey D’s can afford the Hamlet-like hesitation about the “perfect fat texture of their product”. Mickey D’s is looking for success in the aggregate whereas Minetta Tavern is looking for perfection in the particular.

And that ultimately is the difference between prop and rules-based algo.

The biggest problem most traders make is they attempt to turn each Mickey D’s patty into a Minetta tavern masterpiece and that is the road to ruin. Algo trades are not meant to be perfect. That means that they will suck a lot. They will be subject to idiosyncratic wobbles in the market. They will sometimes post six, seven, eight losers in a row – something that would horrify a prop trader. But that is perfectly normal. As long as your quality control remains in place – and for algo trades that means that you’ve created the best possible risk-reward structure and have not overleveraged the account in any way – the long term expectancy will be positive and those pips will pile up.

The one key difference – and this is perhaps the most important point of all – pips profits do not pile up like hamburger profits. There probably no business that has less volatility than a McDonalds franchise. By that I mean, there are probably very few MCD franchises in the world that lose money on a daily basis EVER. That, of course, is totally not the case with algo trading. Pips come in batches and often leave in batches and you have to fully be prepared for a very lumpy distribution.

But in the end – whether you trade prop or algo or both – KNOW the DIFFERENCES – it will help you succeed with each approach.

An Algo for Your Bra and Your Account

Boris Schlossberg

As New York Times recently wrote, these days there is an algorithm for everything – even a bra. Apparently retailers now have a monstrously complex 2000 factor method to make sure that ladies have that perfect fit.

Algos, have been on my mind a lot lately, especially this week after I attended the FXIC forex conference in New York where I had a chance to meet a bevy of algorithmic traders and talk shop for hours on end. When K and I started out in this business we were the classic “have-a -hunch-bet-a-bunch” proprietary traders, but as computers in the retail forex have moved from analytics to execution we find ourselves moving more and more towards an algorithmic way of trading.

Algorithmic trading sounds intimidating, but it is nothing more than modeling price behavior which in turns means understanding how humans react. Although most novice traders believe that algorithms are highly quantitative in nature that’s not so. Algorithmic trading is less about mathematics and more about psychology. In fact many very successful algorithms are ridiculously simple in their premise and exploit one or two common market behaviors with clear, logical risk controls.

Just like in real life anything can be turned into an algo. I am even developing an algo that tries to take advantage of the idiocy of the markets, which I have dubbed “The George Costanza” and for which I hold strong promise.

But perhaps the most interesting takeaway from this week’s conference is that despite the massive array of programming talent, most algo traders I met admitted in private conversation that they maintain strict human control over their creations. In other words, no matter how good your algo – it is at the core just a dumb piece of code that will inevitably miss the nuance of the market.

It is at those times that the truly successful algo traders intervene and take control over their code. Mind you I am not saying that you need to second guess every trade – that is a very common and tragic mistake that many traders make. Rather, I am saying that despite the massive computational power that we all posses, there is still room for art in trading. There is still a time when the most important part of the model is to know when to turn it off.

Like to trade the EUR/USD? So do we!

At BKForex, a large part of our trading is short term and the EUR/USD is one our favorite currency pairs to trade.

The EUR/USD is the world’s most actively traded currency pair and for many forex traders, this activity provides opportunity.

Sign up to receive our FREE 17 Page EUR/USD Trading Strategy

*By signing up for this strategy you agree to receiving promotional emails from