## simple, tradable, awesome statistics

never design a trading system without proper prior analysis. post your market statistics here

Moderator: moderators

rank: 50+ posts
Posts: 56
Joined: Tue Jul 03, 2012 10:08 pm
Reputation: 0
Gender:
Contact:

### simple, tradable, awesome statistics

I wanted to start a thread which boils statistics down to their most simple tradable form. Put specific instruments aside, and lets just talk the plain numbers, their meaning, and practical application.

Any statistics should answer the following questions:

1. what makes this statistic tradable?
2. how does it help capture some of the average range (of the specified TF)

so far the only statistic that makes sense to me is this:

average wick size vs. average body and wick in direction of candle color.

Across different instruments, I keep getting the outcome of:

25% wick/50% body/25% wick

How is this tradable?:

1. after the candle retraces and creates an "average wick" there should be about 3 times as much profit to catch in the other direction.
2. On paper, this should catch up to 75% of the avg range of the candle, with a 3:1 reward/risk ratio

This is my first attempt, and I will be trying to come up with other ways as hard as I can, I truly believe statistics are what creates wins, and I am looking to deepen my understanding and application of them.

Please, if you find any flaw in this, or see anything that is flawed logic, please let me know.

Feel free to post your own findings/research for discussion.

Thank you for your support.

dojirock
rank: 1000+ posts
Posts: 1761
Joined: Tue Nov 17, 2009 10:11 pm
Reputation: 528
Gender:
Contact:
Welcome to the forum!

Good luck!

dojirock
It always takes Momentum to break Momentum!
"A small loss is just as satisfying as a large gain" -MO
"Sometimes we need to stop learning and start thinking...."
"Once you stack, you'll never go back!"

rank: 50+ posts
Posts: 56
Joined: Tue Jul 03, 2012 10:08 pm
Reputation: 0
Gender:
Contact:
I have been trading AAPL, and using a pretty simple strategy lately that has been working well, although it has some pretty big flaws. The strategy is as follows (larger context analysis aside):

1. identify trend with higher highs higher lows, and place conditional orders to catch the trend once it breaks previous highs

Problems:

1. It seems to move against me right away every time in a retracement move before (hopefully) continuing its trend in the given direction. It finally hit me that
this strategy would be much more effective in a situation where AAPL was ranging, and this would catch the range breakout which is more likely to be a very strong move.

2. When I enter, it is usually at a spot where the reward/risk ratio is relatively small, because I will not know if the trend has reversed until the low is broken with a lower low

So this leaves me with the dilemma of what to do when the market is trending, and I guess I have to enter not on the newer high, but when the retracement ends and the trend resumes

PROPOSED METHOD:

from my data, AAPL has a D1 average range of about \$10, and an H1 average range of about \$4, I am looking to capture at least a \$2 stock move every trade (trading options) so this would
place my target holding time at about .5hour - 1 hour.

Now the average wick (retracement) of an H1 candle is about \$.90, and this is consistent with my everyday experience watching the chart. So when AAPL is trending I am looking to enter after
an approximate \$1 retracement with a stop loss at about 1 standard deviation from that (about \$.60) which I am hoping happens to coincide with the previous low that was set as well.

This should effectively move my risk from (retracement plus standard deviation) to just the standard deviation itself

heres a picture to explain:

I know it seems fairly obvious that one should enter after retracement ends and trend resumes, but it is not always easy to pick the right spot to do this, I am hoping that being armed with these few simple statistics will give me the edge I need to enter at the right spot when it is trending!

Any opinions, or criticisms are welcome, I would like some feedback if you have it. I hope my post made sense, my mind wanders a bit when writing, so please let me know if anything is unclear.

rank: 50+ posts
Posts: 56
Joined: Tue Jul 03, 2012 10:08 pm
Reputation: 0
Gender:
Contact:
I found a perfect instance of this hapenning today, and I just wanted to share it as well because it fell right into what my stats say.

EDIT: All in all my stop is tighter, my risk is lower and I have gained an extra \$1 I would have otherwise lost out on.

I think 1 standard deviation on the retracement should be enough, although only time will tell if I keep getting stopped out.
Last edited by deltaskelta on Tue Jul 17, 2012 10:59 pm, edited 1 time in total.

TheRumpledOne
rank: 10000+ posts
Posts: 12947
Joined: Sun May 14, 2006 9:31 pm
Reputation: 1610
Location: Oregon
Real name: Avery T. Horton, Jr.
Gender:
Contact:

Do you post on BabyPips under the same ID?
IT'S NOT WHAT YOU TRADE, IT'S HOW YOU TRADE IT!

Please do NOT PM me with trading or coding questions, post them in a thread.

Thank you for your support.

rank: 50+ posts
Posts: 56
Joined: Tue Jul 03, 2012 10:08 pm
Reputation: 0
Gender:
Contact:
I do, the post from relativity over there got me thinking in a new way, and I'm trying to derive my own methods from the same concepts

why do you ask?

EDIT: TRO, I have been reading through your posts on different forums, and trying to digest everything. I just find it easier to try and derive some of this stuff myself to arrive at my own conclusions which I think are much more powerful than trying to understand someone else's.

But thanks for all the work you have done, and indicators you have released, I am more into equities than forex so I don't have the chance to use many of them because I have to trade outside MT4.

TheRumpledOne
rank: 10000+ posts
Posts: 12947
Joined: Sun May 14, 2006 9:31 pm
Reputation: 1610
Location: Oregon
Real name: Avery T. Horton, Jr.
Gender:
Contact:
I asked because I was curious. Usually, if someone from BabyPips comes over here, they are up to no good...LOL!

I wrote hundreds of indicators for TradeStation and eSignal.

You're welcome.
IT'S NOT WHAT YOU TRADE, IT'S HOW YOU TRADE IT!

Please do NOT PM me with trading or coding questions, post them in a thread.

rank: 50+ posts
Posts: 56
Joined: Tue Jul 03, 2012 10:08 pm
Reputation: 0
Gender:
Contact:
I think I better take a step back here and explain how I see statistics and their use....

1st (LEVEL 1).............How is a "dry (decisionless)" casino game run? (I'll make a simple new game)

-you try to guess the outcome of a dice roll, the dice has 5 wrong picks and one good pick (5:1)
-the casino is paying you 4:1 on your money for guessing the correct
-because there are more possible outcomes than you are being paid for, the casino will always win, beating the game is an illusion, IT WILL NOT HAPPEN
-this game is a simplified version of roulette

2nd (LEVEL 2).............How is a "moist" (minor decisions) game run?

-This would be like blackjack, the player has some decisions to make which will increase their odds, but never to the point of positive expected value
-Playing blackjack within the confines of the dice game above, you may be able to lower the house odds from 5:1 to something like 4.5:1 through proper blackjack playing
-There are some decisions to be made by the player, although the astute player will realize that there is always one proper decision to be made, and one can only lose more money by making wrong decisions, you cannot win more money by making right decisions.

3rd (LEVEL 3).............A "wet" (major decisions to be made) game

-This game would be like playing poker, the house odds are not set, the outcomes are "random" like price action
-It is your responsibility as the player to take these seemingly "random" events and make order out of them, and then dynamically plotting this order against you opponent to win
-Both sides are free to make decisions in their best interest
-This type of game is most similar to trading, I hope you can see the similarities

I can only see how to apply trading to the first type of game at this point, so I will try to reach a level 3 game as my understanding and this thread progresses.

LEVEL 1

1. We have to treat trading like the simple dice game above, only we do not know that there are 6 sides to the dice. The dice roller is behind a curtain, and he just calls out which number hit.
so how do we find out the odds? The simple answer is you cannot ever know for sure how many sides there are to the dice

2. You can, however, make a large number of observations to see how often each number hits, and after your sample size gets big enough, one can predict the odds well enough. (after witnessing 10,000 rolls, my money is going to bet that each number will hit about 1/6 of the time)

3. This is how statistics must be used in trading, from knowing what has hapenned over a large number of observations, we get to know how the instrument (dice, or stock, or currency pair) behaves over time, and most likely how it will behave in the future although there is no guarantee

4. Since there is no "house" and there is only other traders, we are free to choose the side we want, so we must choose to be the house and have other traders be the players, and therefore we have the edge.

More to come......

rank: 50+ posts
Posts: 56
Joined: Tue Jul 03, 2012 10:08 pm
Reputation: 0
Gender:
Contact:
FYI, I started this thread to get my ideas out and on paper, and hopefully it will force me to make reasonable conclusions that I otherwise would not be forced to make, and if anyone cares to add anything, thats just a bonus for me

Anyway, the 5:1 dice roll can be looked at as risk, while the 4:1 payout is the reward, so we have a random event where the risk outweighs the reward AKA the house wins. But I am going to try and compare this to how we can gamble on price action

this is what I know for AAPL:

1. mean D1 range = \$10 stdev = \$5.5
2. mean D1 body = \$5.3 stdev = \$4.8

Now since the mean range is 10, and the mean body is 5, that means that at some point on the average day, the body must be somewhere near \$10 in order to cover the \$10 range, which also happens to be about the stdev of the body
So, when the D1 body is in the ballpark of \$10, it is most likely reaching an extreme, and we can probably take the opposite direction since 80% of the time (4:1) it will close within +1 stdev or less from where it is.

We have 4 good outcomes and one bad outcome, and we can therefore afford to payout at a level of anything less than 4:1 (3.5:1, 2:1, 3.9:1) and still be profitable.

There are a few places where this breaks down in real life:

1. trading is a level 3 game, and this, so far, is level 1 (the payouts are not set in stone like in a casino. While I can define my payouts in the form of SL, there is no way I can define the size of my wins)

-Therefore, I suspect that in practice, one would have to drastically lower the amont they are risking in order to make up for the "risk" of not knowing how big their win is going to be, just knowing that 80% of the time I am going to win "something"