Yeah, I've noticed the limitations in TS also. I tried to codify my method for more than a month and found I couldn't get it to accurately reflect the trades. I tried lots of code like yours and just kept missing most of the trades. I guess it can't be done. Thanks for your effort.
you said you were very sure no pm's sys would lose mony "BIG TIME" over time as proven by your "team" of programmer is what you said when you entered this journal.
You obviously misunderstood my post about not knowing. I've tried to even turn the systemized approach to subjective approach is my intentions but anyway.. I believe from previous posts No_PM suggested the accuracy rate has been 70%. This is what I am trying to prove and for the life of me can't see it. From previous posts, a trade was exited on the stop or the profit objective, and no_pm tried other methods but they proved to lose money. (ie.. breakeven, time stop, etc..) So it is safe to assume the methodologies success rate is determined by a) the ability for the adx (the whole trade setup cannot happen without adx > 30) to identify when the market will trade from one side to the other and b) in conjunction, the oscillator identifying the turn. If there are on average 20 occassions per month when ADX is > 30 then the setup begins. If only 40% of the setups on average trade from one side to the other that means on average 8 can be PROFITABLE trades. From those 8, no_pm suggests that 70% are accurate meaning 5 .6 trades will be identified by the chaiken oscillator. (This is in my opinion the key. Can the oscillator provide the edge?) We haven't even taken into account the other 60% of times ADX is greater than 30 and Chaiken signals divergence leading to stopped out trades. The code was meant to identify the methodology and give a basis for the results to be studied to see the 70% accuracy. Wish no_pm well in his trading.
One thing I noticed early on is NPP is watching the big S&P contract data, but trading on the ES. I had never noticed until then that the two contracts could differ in new highs/lows & divergence enough to affect a system. In all the system testing being done are folks using the big contract to setup/trigger and executing on the mini? Could it make a difference?
I can't give you hard numbers because I've not been able to create my method in TS. My OPINION is that the ES volume is unreliable using my method. My rationalization is this: The volume divergence I'm trying to detect is caused by institutional players. The guys that buy a SP contract, then buy shares at the NYSE, and then exit their futures position are what I'm trying to monitor. These guys are hedging using futures to reduce slippage when they put on their positions. When they decide the market is outside their buy zone, they'll stop buying futures first. This drop in volume tells me they no longer want to establish positions in the market. I believe nearly all of the big money trades are done in the pit. I also believe the Emini is predominated by smaller traders with no desire to actively hedge positions. I believe most of the volume in the Emini are traders bouncing in and out of positions based on whatever speculative reason they have for putting on a trade. I also believe some of the volume is caused by arbitrage between the pit and the Emini. All the Emini volume combined then is more reflective of the level of speculation rather than the level of hedging. This difference is my guess as to why the Emini has different results with this method. All or part of this could be completely wrong. If it is, then the method should work equally well using either datasource. My own testing and trading has indicated that I need to use the SP data to obtain reliable signals.
No trades today for this method. Almost had one with the upmove at 11:15. Unfortunately it didn't set a high above 1003 so there wasn't divergence between price and volume. Hopefully better luck tomorrow.
I appreciate no_pm for sharing his method by starting this thread. Thank you. You've detailed several ingredients essential to a solid trading plan. However, I disagree on the volume analysis of the pit SP vs the ES. here's the facts: the CME sends trade size (volume) for each ES trade because it's executed by Globex. They do NOT send any type of per trade size (or volume) on trades executed in the pit. All they report is end of day volume totals for pit traded contracts. The Chaikin accum/diss calculations rely on 'volume' information. Since intraday volume on SP does not exist, then I believe that the calculations are performed using tick volume for SP. The ES however, does provide trade size, so you're comparing apples to oranges. The 'institutional size' is not being analyzed at all by looking at SP, since it's not even reported intraday. The tick volume on SP is also misleading, because the CME only sends price CHANGES on pit contracts, so each SP tick is different price. Globex reports ALL trades, so you'll see many ticks at the same price. Look at a tick-by-tick chart and you'll see what I mean. The SP tick chart will be all angular as it bounces up and down from each tick, while an ES tick chart will flat line while trades kick off repeatedly at the same price. So be careful what you 'think' you're looking at.....but hey, if it works, then don't overanalyze it.