Now WhyNot is trading relatively small size (less than 20 contracts) on the @US which has volume 100’s of times or more the the DJ. he is using stops and limits. The stops are effectively market orders after their price is hit. Limits that are penetrated should have 100% realism.
Sam, Can you explain me why this trade got a ZERO Overall Realism Factor ?
BTO (STOP) 17 @USM6 107 15/32 5/19/06 - STC 17 107 25/32 5/22/06 (LIMIT- Price reached 107 26/32)
I really dont understand, what is the magical alchemy?
Well I just glanced at the bid volume / ask volume at a random moment (the moment I read your post) and it was bid size 30. Perhaps that was a bad time to look. I’ll do further analysis.
I appreciate that you are trying to make the formula as accurate as possible. That is my goal, too. We’ll get there (eventually).
And Mario - Your points are well taken, too.
Matthew
From Investopedia.com "Market vs. Limit
The two basic types of orders that every investor should be aware of are the market order and the limit order.
A market order is an order to buy or sell immediately at the best available price. These orders do not guarantee a price, but they do guarantee the order’s immediate execution."
Sam, why would it take 1 hour for an order that must be immediately executed specially on a measly 200 * 5 contracts on the mighty Dow. Mind you, you will get a disadvantaged fill but it would be immediate, that is why they call them market orders. Unless the market is electronic, the reported fill may perhaps take an hour or so.
Matthew: If I can give you a suggestion, you would have to remove "autotrade factor" from RF average per trade because it is too much variable (TradeBullet works?Run?Bugged?Trader is sleeping?He is on holiday etc…).
You after could add a value outside the calculation and name it "TradeBullet Factor".
It is only my suggestion
I get very reluctant to throw my $.02 in these arguments, but I think Fanus has a point. He is trading some of the most liquid markets in existance, yet he scores a 0% RF? Now, mind you, I don’t know what order type he is using (limit?, stop?, market?), but in the markets he is using this seems a bit harsh…
Hans. (now back into lurk mode…)
"Sam, Can you explain me why this trade got a ZERO Overall Realism Factor ?"
No. I agree your stop orders should be near 100% and the limits not far behind. It makes no sense that small orders in high volume markets are rated nearly the same as large orders in very thin markets.
"Sam, why would it take 1 hour for an order that must be immediately executed specially on a measly 200 * 5 contracts on the mighty Dow. Mind you, you will get a disadvantaged fill but it would be immediate, that is why they call them market orders. Unless the market is electronic, the reported fill may perhaps take an hour or so.“
Look at the volume. The “mighty Dow”, as in DJM6, often has fewer than 200 trades in an hour. A market order of 200 x 5 subscribers wouldn’t get filled in an orderly fashion. Just look at the charts or time and sales.
My time and sales shows under 1000 1 lots ALL day today in DJM6. What does your time and sales data show?
Conversly I show 1000’s of 100+++ lot trades in the ZBM6.
Anyway, I don’t know if it would take hours to fill a 1000 lot in the DJM6, but I do know a market order that size in the “mighty” DJM6 would kill most retail accounts in slippage costs alone.
” From Investopedia.com "Market vs. Limit "
Would you like to talk to a floor broker or a book? DJM6 is the pit traded Dow. Not the Dow stocks, not the eDow (YM). Let me know. I’ll get you in touch with my guys down in the pit…they’d love to fill your 200 x 5 DJM6 market orders.
Fanus, I know which orders you are talking about. I can tell you that the Bond and Note orders from your system are not sent right away to tradebullet, but hours later, even when the session has started already. I remember even to have seen that more than once.
When I enter new orders in my systems (riva balanced and seasonal), I receive confirmations from C2 10 to 30 minutes later. This was also the case for the Eagle One and Black Bird Trading program, whereas the Aestreux Funds signals come through right away.
That is simply not correct.
For instruments that do not trade 24 hours, if an order is entered during a market-closed period, then the C2 AutoTrade server will hold the orders until a few minutes before the markets open. This is a design decision - not a mistake - since several brokers will reject orders sent too "early." I am unaware of any cases that C2 simply randomly decides to hold orders for a few hours. If you have a specific instance of this, then contact me with the details. But I do not think it is correct.
"Sam, Can you explain me why this trade got a ZERO Overall Realism Factor ?"
FWIW, I apologize for my first post. I was doing other stuff and realize my point isn’t that clear. Let me start again. I commend Mathew for C2 in general and the RF attempt in particular. Hopefully my real life trading experience can help in the evolution of the RF. I see some unfair inconsistencies. Here is an example:
WhyNot BTO 17 @USM6 107 15/32 5/19/06 11:11
STC 17 107 25/32 5/22/06 11:52
That trade has a RF of zero. Volume in the @US was 10’s of times the trade size at the time of the trade. The RF should be 95% or higher.
Test System STO 200 DJM6 11563 5/5/06 11:29
BTC 200 11624 5/8/06 9:45
That trade has a RF of 100%. Volume in the DJ was less than 10% of the trade size. Not sure how low is right, but you can bet there would have been HUGE slippage on this fill even though it is a market order.
RF of 5-20% is possible. The order could even be rejected by the floor. Very likely you would get a call from the floor manager asking you if you know what the heck you are doing. The bottom line is 10 or 20 subscribers trading 200 DJM6’s each would create more volume than the DJM6 has ALL DAY LONG, while 10 or 20 subscribers trading 20 @USM6 could be absorbed in a single @US tick.
Another observation is that it seems @ MKT orders get a free ride in the RF. In reality there is a delay even in auto trades between a vendor’s sending a market order and customer fills. That delay is much larger on IM or email executions. OTOH, MIT’s or Stop orders are by definition market orders once triggered. Moreover, they are waiting in the market and should be executed without any transmission delay. In short, MIT and Stop orders in liquid electronic markets should have the best RF.
This isn’t based on a definition from some book, but rather 1000’s of real life trades.
Last, IMHO, systems that consistently abuse RF criteria and margins should be punished in the rankings and “best system” listings.
Systems like WhyNot and others that are “real” should be rewarded.
I think we have demonstrated enough that RF dont work and it is basically "TradeBullet Factor" We hope this discussion will not die on this forum page as many other thing asked from system developers because IMHO our interests are the same as yours and also it would be more clear for all potential subs.
Matthew
Are you still investigating the Realism Factor? In my Snapback system for the last 4 orders, I received a realism of 0%. And that for 1 contract in 10 Year Notes with stop orders, 2 contracts in US Bonds with stop orders, 3 contracts in @NQM5 with market orders and 4 contracts in @ESM5 also market orders.
If it continue like this, the system will have close to 0% realism pretty soon even though it trade some of the highest volume futures contracts on stop and market orders.
RF is cleary still in a development stage and not ready for general usage. Until the issues are resolved, I would prefer it be removed from the statistics pages, or at least stop including Autorade factor to calculate this. As Mario has said before the way it is now, it is really an Autotrade factor and not a realism factor, since if someone’s autotrading is not working for whatever reason, the realism factor is 0%.
Can you please provide me with the log files from C2/TB where it show the orders was sent to the broker and the broker came back and say it was not able to fill the order? Since you use this in your calculation for RF I assume you must have that information available.
Regards
- Fanus
As Mario has said before the way it is now, it is really an Autotrade factor and not a realism factor, since if someone’s autotrading is not working for whatever reason, the realism factor is 0%.
Fanus: More exactly it is “TradeBullet” Factor, and not “autotrading factor” because, outside C2, autotrading with software have execution on the server is almost perfect 
Ciao
Matthew
Here we go again. Latest trade in SnapBack system.
Close 4 contracts of @EMDM6 with a market order at 9:30.
Volume factor - 100%
Stop and Limit Factor - 100%
Autotrade Factor - 0%
Overall Realism - 0%.
About a 1000 Contracts were traded in the first 3 minutes. A market order with 4 contracts and for any subscriber easily should have been filled. I didn’t have any problems getting my trade filled. Mine was just filled at 9:30 and not a minute late like on C2.
All the realism factor is telling, is that C2’s Autotrading option is not working at all. This is a C2 problem and not a system problem. Please do not penalize the system for something C2 is not able to do.
I asked for it yesterday and I ask again. Please show me the log files where it show that C2 sent the order to Tradebullet, Tradebullet sent it to the broker and the broker come back saying it was not able to fill the market order. You must have this information, otherwise how else can you calculate the Realism Factor?
Ignoring an issue is not going to make it go away.
Regards
- Fanus
Fanus:
I’m not ignoring your issue.
I said I will get to it, and I will.
I can’t produce “log files” from TradeBullet. That’s not how the number is calculated. The way it works is: C2 has a database of orders sent to various AutoTrading clients (TradeBullet is just one). It keeps track of how many of those orders are filled. That percentage (i.e. number of trades filled in brokerage accounts versus number filled in C2’s hypothetical engine) factors into part of the Realism Factor (RF).
There’s no need to argue. I agree with your (and Mario’s) complaint that having a Realism Factor of zero in cases where (for example) a single user is sent an order, but forgets to turn on his TradeBullet, is not very useful.
I will try to fix it soon. But there are other things I do here, too. I’d like to think that, while I’m not as fast as my customers would like, I am at least a bit more responsive and interactive than most software companies. Please hang in there.
I’m working as fast as I can. Realistically, maybe I can get to it this weekend, but I’m hesitant to promise a date, because I don’t know yet for certain.
MK
Matthew
The best thing to do while you are fixing the problem is to disable the calculation of RF. While you are looking into the issue, RF for many systems is unfairly getting trashed in the mean time.
Also it shouldn’t just look at how many orders were filled. It should look at how many attempted orders were filled. If someone’s auotrading software was turned off, that should not be counted at all. Only the orders sent to the broker from the autotrading software and not getting filled should be counted.
- Fanus
"I’m not ignoring your issue.
I said I will get to it, and I will."
Matthew,
I think your responsiveness and non-defensive attitude is great.
Thanks!
Here’s some ideas and an observation on the RF. Let’s look at
two great systems: Aestreux Fund and WhyNot.
Aestreux Fund: Cumu $ $464,587
and real-life slippage $389,756
Realism Factor 93.6
WhyNot: Cumu $ $109,150
and real-life slippage $101,268
Realism Factor 82.0
So, just based on the present C2 stats the Aestreux
RF should be about 84% ($389,756 / $464,587),
while WhyNot should be near 93% ($101,268 / $109,150).
It’s ironic that they are nearly exactly the opposite.
I’m sure there is a more complicated way to do this, but
having an IQ of less than 200 I like to keep things simple.
Am I missing something, or does this make sense?
Thanks again.
Sam: As you can see "WhyNot" TradeBullet Factor (aka RF) is coming down. Every time we have a great trade TB Factor comes down, it is really funny
My broker and other traders here following "WhyNot" was amazed because this system works with one of the most liquid market, entry order are always stop order and the target is often (not always) filled crossing the limit price.
Real "Realism Factor" for "WhyNot" would have to be 90/99% circa (not 100%, I agree, because we can have rare slippage and sometimes target are filled only touching limit price).
Actual TradeBullet Factor is totally no sense (also you have to remember that TB excute order on the client).
We must ask Matthew to temporarily cancel RF from the statistic until he fix his algorithm or, as I have already suggested, he would have to change the name in TradeBullet Factor.
I am insisting on it because many potential subs are scare from rf for no reason and it damages me, Matthew (he clearly gains from subs) and all C2 community because the truth are totally distorted from that parameter.
>Sam: As you can see “WhyNot” TradeBullet Factor (aka RF) is coming down. Every time we have a great trade TB Factor comes down, it is really funny
It is “funny” because the RF based on C2’s “actual data” should be 93,
but:
Realism Factor 73.8
Cumu $ $112,520
and real-life slippage $104,487
If there is data on “real-life slippage” then why doesn’t C2 use that?
If there isn’t data on “real-life slippage” shouldn’t that be a red flag?
Why is a system that is actually trading punished relative to systems
that are not?
The determining factor for RF on systems with subscribers, is tradebullet. If any subscriber turn off tradebullet for the day for some reason, C2 assumes Tradebullet cannot fill the order and give the trade a 0 RF.
It does use slippage too, but multiply it with the RF value for Autotrading. So, it doesn’t matter how good or bad slippage is. It gets multiplied with the Autotrading factor and if Tradebullet is turned off, that get a value of 0, so the resulting RF is 0 as well.
- Fanus
