Sunday, March 30, 2014

Why Don't I Trade My Plan When I Plan My Trade?

Nothing is quite so frustrating as planning a trade, not following the plan, and then seeing it work out--without you on board.

Why does that happen?

Kahneman's work on fast and slow thinking suggests that we have two information processing systems at work at all times.  The fast system is highly efficient and responds to immediate situations; it processes information automatically.  The slow system is effortful; it reflects, analyzes, and reasons.  If someone gave you a challenging math problem, the fast system might respond with, "Yuck!  I hate math!", while the slow system would go to work breaking the problem down.

When we plan an action, we engage our slow processing system.

When we are in the heat of battle, our fast processing kicks in.

Indeed, that fast processing can kick in when we are least aware, creating perceptual biases and shaping our behaviors.  

As Mike Tyson famously said, "Everyone has a plan until they get punched in the mouth."

Harder planning, better planning, more planning--none of that will work if we are in a different mind when it comes time to execute the plan.  That is why special forces teams rehearse maneuvers under simulated battle conditions; why pilots practice on realistic simulators rather than in an empty classroom.  The great advantage of mental rehearsal is that you can imagine challenging situations--and evoke the reactions of your fast system--while walking yourself through the right responses.  

Either you battle test your plans in rehearsal mode or they will be tested in actual battle.

Further Reading:  Visualization Techniques for Traders 

3 comments:

Curtis said...

Key insights:

* It is harder to execute your plan when it is right then when its wrong.

* Imagine a trader is bullish on the market and plans to buy near the prior close. But, the market doesn't open at the previous close but instead gaps up. The trader now experiences cognitive dissonance. The best plan was too buy the market at the prior close but that's not an option. The trader can now either wait for the market to come back down to the prior close which will reduce the probability of the trade working out, trade long from a higher risk position, or go short from a lower risk but lower probability position.

* In the above example, if the proper move is now to short the market -- the trader who relies on plans will experience cognitive dissonance "I said I was going long but now I'm short." And without strong training is unlikely to be able to hold the short. If the trader waits and the market rallies, he is likely too attribute it to not executing the plan "My plan was to get long.". Should the trader go long and get stopped out, it will be attributed to not having patience. There are techniques to minimize such problems.

* One experimental but highly promising technique I've used, I've called "Hammer style". This trading technique relies on being both long and short the same market at the same time. Obviously this is technically flat and unfortunately not directly supported in futures but can be simulated using software such as NinjaTrader. The power from this technique comes from the ability of trader to remember their position across multiple strategies and to multiplex multiple strategies across the same instrument. The power from the technique is that it opens up many more possibilities too win.

* The market tends too make the biggest moves from the greatest times of uncertainty. I have tried many techniques for avoiding getting faked out of a trade at the worst point, including adding random timers to my trades. But, the fact is the market is most likely to "decide" when indecision is highest.

* There is an information complexity increase from knowing where the market is going to do too actually placing the trades.

* If even very accurate trades are bounded too precisely then the result will be consistent losing. Most losses are caused by too precise bounding. Due to random variability, it is impossible to know the proper bounding on instruments like futures. Binary options which "volatility normalize" the market thus have the potential to be quite powerful.

* Again, I've presented two suggestions for bounding losses strategies. The first was too multiplex multiple strategies over the same market. The second was too use a risk limited instrument.

* Another technique is to move toward an event-based market processing style which minimizes the importance of plans. We call this "process based trading" and is a specialized form of trading.

* Tactics "implementation" or "logistics" often requires different competing mental processes that can easily take a trader out of flow. It might help to have one trader plan the trade and another execute the plan.

Curtis said...

Thinking some more of new paradigms in trading...

Let's imagine how the multiplexing trader might approach the scenario though of the market gaps up with a long bias provided he can trade the same instrument long/short and set the point multiplier.

Long ES1 @ $20 per point. Start of day bias is long.

From a process oriented or event-based approach, the trader spots a short opportunity.

Short ES2 @ $50 per point. Closes for $100 profits.
Short ES2 @ $50 per point. Closes for $80 profits.
Short ES2 @ $50 per point. Closes for $50 profits.

The trader has closed $230 trades. And now has let's say $100 in open losses. It can be looked at in many ways, the trader has bought himself a better entry, the trader has retained memory of the original plan, that the trader has spread instrument against itself, or that the trader has managed to gain greater profits from horizontal movement then vertical movement. The closed profits act as hedge and allow for profiting in many types of markets except for markets that trend strongly.

One of my ideas for this style came from a realization that every prediction that I make has an edge. So, you'd want to take every prediction right.

Alternative possibility, the trader comes into the market with a long bias and thus buys a binary option that pays out should the market rally. The trader then focuses on generating closed profits on the short side. The benefit of this technique is that there isn't as much pressure to generate closed profits -- as in the previous example-- the pressure would be very high if the market started to trend strongly against one, somewhat mitigated in the example above by our ability to set the point multiplier.

Curtis said...

Summary of common problem scenarios and suggested solutions

A. Problem scenario: A day trader ,who focuses on the cash session only, forms a bias that he'd rather trade long from the previous day's market activity. However, the market gaps up before the open to his target.

Solution:
We can imagine that many day traders will experience a similar phenomena. The highly trained process based trader with real-time event based processing is more likely to use the plan as just a rough guide and formulate new hypothesis from developing information. If a trader is thus struggling with this too much then it is likely because they haven't learned the specialized skills required to actively day trade. Rather they are a swing trader operating on the day time frame. The solution for that trader is the recognition that his primary advantage came from the prior day's session and to place his trade earlier.


B. Problem scenario: A trader has the right plan but frequently finds that due to software problems he misses trades, ends up in wrong positions, etc.

Solution:
The problem in this case is very likely using inappropriate buggy software. Changing to a better and more suitable software is the solution.

C. Problem scenario: The trader has a good plan that calls for risking $x on the trade. However, instead he risk less then $x and loses the trade.

Solution:
In this case, either the trader did not accept the full risk that the trade required for or feels that he/she can use real-time information to achieve a better result. If the latter then it is not really a problem but rather an open conjecture-- if the former the trader must accept the risk upfront.

D. Problem scenario: The trader generally has very accurate calls on the market but yet loses frequently.

Solution:
This problem is typically caused by bounding the precision too tightly by using very restrictive time limits or stop constraints. When precision on the trade is tighter then the accuracy then the result will be losses. The solution in this case is not easy because the optimal bounding is unknown or too large to be applicable. But, some solutions would be too find a better bounding, move from futures to options (tight to lose bounding), or to structure the trades using more information, i.e the trader determines the direction but uses algorithms to define entries.

E. Problem scenario: The trader's core plan radically shifts from the planning stage to the implementation stage. The plan called for a bullish position but the strategy utilized was a neutral options strategy.

Solution:
In this case, the trader is unlikely to be able to prioritize risk appropriately or has a strong aversion to taking directional risk. During the planning stage, the trader evaluated everything objectively but during the implementation stage, the trader took on the hat of not losing any money. The result is a position not aligned with the original strategy. The solution is for the trader to keep a keen focus on the core strategy.

F. Problem scenario: The trader finds that he is placing many small trades over a short period of time and losing more money then expected.

Solution:
In this case, there is a cognitive dissonance likely caused by the price that the trader sees versus the price that the trader gets. As the trader attempts to bound the losses tightly, the impact of slippage and market orders become more pronounced. Ironically, this traders problem is both a case of over trading the same idea and likely under trading other good ideas throughout the day. The solution is for the trader to reduce the usage of market orders and reduce the number of trades taken per the same idea while increasing the number of trades taken for other ideas. Alternatively, the trader can increase the risk taken per trade which will have a similar effect. This problem is similar to the cause of other bounding problems.