It looks like that the large position on the
ERY and the money locked in the CL PUT 54 are in practice making arduous to recover, with the available funds. In fact, even the favorable moves end up generating further losses because the lack of funds do not allow to close the position created for hedging on the
ERX layer. This is, in practice, causing a simple "transfer" of the loss from an instrument to another (in this case from ERY to ERX).
We can note here how the large ERY position is an example of violation of both the "principles" we have been mentioning in the previous post:
- 1. Violates what we have called the "lump of money principle" because it's too much money allocated at one price;
- 2. Violates the idea of scalping focusing on the "horizontal" moves, because a large sum on one spot has mostly directional consequences.
So, this is a "double-size mistake
A chunk of capital used in this way absorbs way too many useful resources which could be used for scalping, and
it forces to take undue losses anytime the price goes against you. In addition, the same happens even if it goes in the right direction, because there may be not sufficient money to close the position on reversal on the hedging instrument. This way, in practice one is digging a ditch and burying himself in it
This situation here is a bit extreme, but, in a smaller scale, it actually happens anytime we are trying to be after directional moves. For this reason, I believe that a "
nonpredictive" approach is the one most reasonable, as it allows to
look at any price move as a sequence of (possibly nested) horizontal levels, where there is a lot of money to be made on "fluctuations". Of course, on top of it one can always apply possible predictive models or personal expectations, in the form of discretionary actions, use of "biases", and so on.
A
purely directional view is instead imho counterproductive because you either have to use a very small fraction of your capital in directional bet (with essentially random outcome: PNL fluctuating around 0 + negative drift due to trading expenses) or, with larger bets, it tends to keep locked money in positions which, even when work in your direction, may still be able to create problems and one ends up not even making money on it. So it's important to "scatter" the allocation over the price range and operate in a dynamic fashion, getting in and out, with a continuous action of hedging on the "horizontal corridors".
It's clearly a shift of vision that in most cases is not very intuitive, because most of the "conditioning" (even from a lot of misleading literature
) we have is to look at the market trying to spot directional moves ("trends"). This thread I think illustrates in practice pretty well this point. If we look at what is going on in the various layers, we can see for instance that
I have closed in profit all the CL option layers (except, of course, the untradable one, CL PUT 54) and the main problem is being represented by a "
lump of money"
allocated on ERY, which is not being scalped but it's freely fluctuating with the market like a big hammer swinging on our head.
That money could have been used much more efficiently in scalping activity with smaller position and little risk, and not being forced to take innumerable losses and liquidations.
So, I think this is pretty illustrative of the suitability of the approach based on non-predictive scalping/hedging, where
we look at each price move as a sequence of horizontal steps, which are possibly "nested". "Nested", in this case, means something like having a choppy market around some price level, then having a directional move down (for instance) and a choppy mkt at that level, and finally returning at the previous level above. In this case, we have 1 "horizontal moves ("first level") and 3 horizontal moves ("second level") which are nested in the previous level (clearly, a highly profitable configuration in our approach):
Instead, these big "lumps of money" left fluctuating with the price direction are only source of problems and a very inefficient way to use the available funds.
Breaking the "lump of money"
Yesterday, reflecting on the "lump of money" issue, I focused
another idea to deal with these situations, which I think can also be of general usefulness in a lot of situations. In the case of ERY, the
big position has been created by me by opening a ridiculously large position on ERY options, and then moving the assigned shares on a manual layer. However, there are other cases where there might be an undue accumulation of money at one spot for various reasons.
Examples of these reasons are the following:
- A
pre-existing large position which you are
importing within the automated application ("position projection")
- A reopen order of a
rollover, transformed into an automated player
- A manual order that you placed either by error, or because you wanted to be greedy and that you want to pass to the automated management
- A
liquidation made by the broker that you have transformed into a player
[etc]
Whatever is the reason, a large player at one spot is always cause of
inefficiency and game disruption. Take for instance the
10.000 shares order on ERY (one resulting from one assignment), if we want the application to start trade automatically on this layer , so that we can start realizing something and not just taking hit with the directional moves, we can of course transform the order into a player (this is done with just 1 click), but the problem is that 1 player of size 10.000 is certainly going to create disruption in the automation, because its abnormal size.
Now I focused a solution to this general problem. This is actually something which has been in the back of my mind for a while, but with the issues of this test I have been finally focusing on it.
What we can do is to take any player which is "fat" and blatantly oversized and
break it into pieces I am aware the idea can be a bit mind blowing when you first hear it, but I can envision it will be a useful feature. Now,
what do I mean by "breaking" a player (open order) into pieces ?
Well, I mean that I take the order and substitute, in place of it, say N
smaller orders in such a way to keep both the
global position and global average unchanged. For instance, assume you have create a big player, say of size n. The player is sitting there with its huge size and it's a problem to deal with it. Then we say, ok let's break it into pieces,
each piece not larger than s shares and let's
distribute the resulting players over the price range, appropriately
spaced (for instance a x% of the avg fill price of the original player).
So, for instance, what we can do is to create a number N of players equal to the ceiling of ( n / s ) and we allocate in the N-1 player the size s and in the N-th player either the size s, if n is divisible by, s or the remainder of the n/s division.
Then, we take these N players and assign them an avg fill price in such a way that if we take all the players together, their combined (weighted) average is the same as the original player. This sounds like a pretty powerful idea, and on Monday we will try to apply it to the ERY layer so that we can take the entire layer and pass it to the bot instead of letting the huge position fluctuate with the instrument:
As a practical example of this, look at the following preview (momentarily connected to the "
demo" account, not paper trading). Let's take ERY and let's break it into pieces the large
10.000 shares player (I added a split feature with a small dialog to specify the
size of the pieces and the
distance between them). We start from this situation with the huge size players, and then we apply the splitting:
thus obtaining something like this:
And same thing for the other big player (
18.100 shares).
Similarly, we can
break also all the other players on any other layers which we feel are a bit large.
Note also the "shift in logic"
Looking at the split of the big player on
ERY layer, it is also apparent that doing this, it also induces a noticeable change of the recovery "logic". You would remember that previously I had placed a few "enqueued orders" for later automatic execution (this was a, say, "semiautomated" solution). Those enqueued orders started from about 21 and when down.
This, however, has caused problems because the fluctuation of that huge position, before arriving to the point where we can start discharging it, can be is very harmful as it wipes out all the available funds.
Now, instead, we are switching to
full automation, and the split players which are created force also a different approach. In fact, what will happen is that the higher players now will tend to remain open unless the price goes up, while those just below the
original player will immediately engage in
automated scalping/hedging, without any further delay, thus possibly freeing resources for horizontal scalping at the various price levels. And actually this shift in logic seems in principle reasonable. Anyway, we will see how it will play out in our situation.
