A byte of (personal) "trading philosophy"
You see now how the concepts of "
Realized" and "
Unrealized" are quite arbitrary, and essentially just useful for accounting purposes (paying taxes).
I understand that some people, who perhaps have not really thought about it and for some reason (possibly just because of "terminology"), have this
naive idea that "
realized" is something that "you already
have in your pocket", while the "
unrealized" is something somehow "fluctuating" in some stochastic world, and that may well be
lost or
regained.
But it is
not so. (What you "have" is just your current PNL.)
So, in essence, when is something "realized"?
Orders can be
matched arbitrarily. An order
can be rematched to another one, as trading proceeds forever. A closed position can be reopened, and vice versa.
What is, then, something "realized", from a
conceptual point of view?
The practical "accounting" answer is obviously that it is what you get by somehow
matching the "lots" (that is,
portions of orders) until you remain with lots that cannot be matched (which will obviously form the
current signed position and the "
unrealized").
The "deeper" ("strategic") answer, for me, is another one, instead.
To me, "Realized" is all that is "
forgotten" or
not used for future trading activity.
"Strategic realization" essentially happens when there is a "
loss" (or a non-use) of past trading information.
This is the main conceptual problem of most "
gambling" approaches. Working "
trade" after "trade", that one thinks is "closed" at a "profit" or "loss" (which is just a mental construct). (When using no trading information, the underlying stochastic process will not allow for any profit of systematic nature.)
Formally, if I had to, I would express this concept in this way, using a
conditional probability:
if
Code:
Pr ( NewOrder | OldOrder ) = Pr ( NewOrder)
we can consider, strategically speaking, the
OldOrder as part of the "
Realized" term.
(This is a purely conceptual, and strategic level, obviously, having
nothing to do with "accounting").
"Realized" is, then, essentially something, that belongs to the past, about which we
disregard the t
rading information, as it in
no way it changes the probability of future orders. (A new order is
independent of any "order" I place in the "realized" term.)
Realizing losses (= losing their "trading information") leads to long-term losing strategies.
A "realized loss" is, essentially, associated with a "loss of information".
(The same applies to a "realized profit".)
This is also the reason why, in my application, all the information about "profitable trades" is
discarded. In contrast, all the information about the "running losses" is accurately stored and actively used,
in probabilistic terms, in future trading decisions, using the so-called "
player" concept (as opposed to a "manual order"), which is essential, an order "equipped" with trading information and automation capabilities associated.
Following this line of thought, in a
strategic sense, the PNL could be decomposed into components:
Code:
PNL = RealizedPositive + RealizedNegative + Unrealized - comms
where:
RealizedPositive = sum of values of all the "players" which closed in profit
RealizedNegative =
0
Unrealized = sum of the values of all the "players" who are waiting for their chance to close in profit
Note that in this definition the "Unrealized" is obviously different from any possible "accounting" definition, because, in our case, the absolute position causing the unrealized, will always be
bigger.
As an example of that, consider the case of just 2 orders on a layer:
One BUY at a high price and a SELL at a low price (same quantities). In this case, an accounting system would just register
this lot matching as a "loss" and the
unrealized component would be
zero.
Strategically, instead, these are distinct "players", waiting for their chance to close in profit. Therefore, the "
strategic unrealized" will take into account both of them and will
not be zero. The corresponding
unrealized position is the sum of the absolute quantity of the two orders, while for any accounting system it would just be
zero (as the two orders cancel out).
Now, since on a given layer, we will have ideally an
infinite sequence of options, and therefore
a price curve constantly going to 0, as time diverges, we are in theory (and practice) guaranteed, given enough time, to be able to close all (except a possibly negligible quantity) the players at a profit, because we can always (given enough time) pick, in the options matrix, a next price curve which will allow us to do so.
Clearly, what we have to
fight against to, while collecting these scalps, are possible
undercapitalization (margin requirements) and volatility and adverse move problems. For the former, we need to make sure we have
enough funds and are not too greedy to use them all, except when really necessary to "survive". For the latter, we use essentially two mechanisms:
-long scalping (of the PUT options) [at layer level] to capture the market correction and volatility burst, with buy stranded player recovery, as explained above
-"protective structures" [at folio level] which have the effect of slightly reducing margin requirements and also provide protection and an additional source of "extraordinary" profits, in case of a "deep" crash or volatility explosion (we will see later what kind of "payoff" shape these structures create, and how it is made "slide" along with the underlying).
Don't forget, my ssh key is here and you promised me a demo on one of the many computers you have. This shouldn't be difficult for a such a wealthy, intelligent, generous and kind person like yourself, right?