Search | User Listing Vidsite | Calendars | Quotes
Home Page ->  Optimizing OmniVest -> Account Administration -> View ThreadLogon (or Register, or Join TradeTight)

You are logged in as a limited-access Guest.To join TradeTight, first read the info in the Organization & Content room, then click the link above. 

John 6:27 (NIV) ... Do not work for food that spoils, but for food that endures to eternal life, which the Son of Man will give you. On him God the Father has placed his seal of approval.


Limit Orders versus MOO/MOC
Jump to page : 1
Now viewing page 1 [50 msgs/pg]
Jump to room :
JimDean
Posted 7/5/2013 10:39 PM (#4973)
Subject: Limit Orders versus MOO/MOC



Owner/Admin

Posts: 3925
2000100050010010010010025
Location:
USA: GA, Lawrenceville
There is a thread in the N forums which discusses this. I posted something there that I think is worth preserving so I'm echoing it here. Click here for the thread in the N forums. The comments below are in response to a request for my opinion on some statistics provided in that thread from test runs done in OT, and some live trade testing of limited duration done using OVest signals for limit vs MOO order types (over different windows of time).

-----

Clearly there are differing viewpoints on this and different ways to interpret the data.

My "educated guess" is that the reason MOO orders have been chosen as the defacto standard is CONVENIENCE, in several areas:

1. From a practical standpoint it is much easier to design and test and prove OVest as a viable product if plenty of TIME is available in which to do the massive batch runs required. MOO (versus MOC) allows a full overnight cycle for that. Wisely so IMHO. Sometimes the processing has been delayed for several extra hours as the kinks get worked out of the system. Once OVest moves to realtime that luxury will be lost. Nirvana was very smart to do it this way.

2. For people who want to trade manually rather than using the TP, providing that overnight window gives them time to review the trades and make decisions if they choose to. My guess is that the foofoorah related to various brokers limitations on when MOO orders could be submitted, combined with higher longitude time zones, was something that may have been initially overlooked but now it's being dealt with using smarter TP logic.

3. Yet another convenience aspect is that MOO order based trades are by far the easiest to simulate and BackTest within the OT world. It's a "firm" price at a known time that all routers and brokers have to conform to. Nirvana wants to be able to test things, and wants their customers to develop confidence. By removing the uncontrollable variable of time and Market-order slippage, they chose a wise entry path. Results are historically verifiable regardless of the source of the data or the path of the execution, if MOO is used.

There are other reasons of less significance but I trust you get my point. I believe the MOO choice was for convenience rather than any scientifically verifiable proof that MOO is the most profitable approach.

Moving on to the other side of the question, re usefulness of limit orders. This moves from the stable, verifiable MOO world into the "mushy" realm of market makers and routing engines and partial fills and so forth. Just because one data feed (broker or retail) says that such and such a bar's price range touched or crossed an arbitrary price point does NOT mean that a viable number of shares were AVAILABLE from broker X using routing Y at time Z, such that all such limit orders would be filled at that time. Way too many variables to make definitive statements. And this btw has nothing to do with the sometimes buggy OT simulation for limit entries.

Look at it this way -
OVest bases its signal to enter on the Closing price. I think it's fair to assume that the chance of the next day's open being higher or lower is roughly 50/50 (if anyone believes otherwise then they should prove it and make a zillion dollars from it). So - logically - if some MOO prices are higher than the prior close and some are lower - ***if you do enough trades*** its reasonable to assume IMHO that the plusses and minuses tend to wash out.

The problem with back testing comparisons is that the sample is limited - and that there is a flaw of some sort in OT's limit entry processing that may or may not be affecting those trades - hard to say since the flaw has not been studied and fixed in the cases that it does occur. It's been documented a lot. Elsewhere and elsewhen.

So, let's just use logic. Presuming MOO's are 50/50 better/worse than the prior close, and presuming the Limit orders are set to the prior Close price, then it's easy to conclude that IF there is NO impact of routing and avail liquidity and so forth that Limit Orders (for trades that were filled), in that perfect world, would produce CONSISTENTLY BETTER ENTRIES than MOO orders, on the whole.

But the specific price of the entry does not a profit story make. Other factors come into play too.

Limit orders also serve to ELIMINATE some trades. Some of those would have been "bad" trades and some may have been "good" trades. No way to tell ahead of time, unless you own a crystal ball. The only generalization I'd care to make is that the SHORTER the duration of the trade (such as RTM trades), the GREATER likelihood that an adverse move overnight would be "predictive" of a loser if the trade played its way out, than for a longer "trend" type trade.

So, in my book, the Limit order approach is the wiser course for quick RTM trades.

Now, since this thread has been looking at a lot of statistics, I suppose I should address that as well (since you asked :-). Please remember that when you do comparisons of "Net Profit" or comparisons of APR, you MUST NOT FORGET COMPOUNDING. The fact that limit orders reject a lot of trades that MOO orders accept means very simple that MORE MONEY IS IN PLAY with the MOO orders than with limit orders, unless you change other factors as well. So, of course, the net profits can be higher - there was more money at risk, after all!

The statistics that need to get more focus is the average percent of account invested. For limit vs MOO runs, all else remaining the same, it's reasonable to see that the average in trade is LESS. So, total profits are less.

And that's where OmniVest's strengths (or soon to be improved strengths) lie. OV's main stated goals include the maximization of use of your funds over time. So, ONCE the response time of OV is brought into realtime, and ONCE there is dynamic updating of Account funds availability from TP back to OV (both things are part of Ed's recent survey and plans afaik), then whatever funds are FREED UP by "intelligent" Limit order filtering, will be RE-useable immediately (or worst case, the next day) for other Limit-filtered opportunities.

This is a GOOD THING. You can make it happen right now if you are diligent about manually updating your Current Account Balance in OVest settings once a day right after the morning run, before the evening run, once you know the net result of which orders were take. And which weren't, and the impact of partial fills. By doing that, you're telling OVest that the Limit orders are leaving more money avail for tomorrow's trades than it would have known about otherwise - therefore it can USE that money.

Afaict all of the discussion up till this time has not allowed for those daily updates (or considered that N is planning to automate that feedback communication from TP to OV). Also, afaict, all discussion prior to this has skipped over analysis of how the compounding effect of funds in trade affect net profit. Proper modelling would call for manual tweaking of allocations when using Likit orders so that, over the average of the test period, the amount in trade using limit orders would be as close as possible to the amount in trade using MOO.

If that procedure is followed and if the occasional inexplicable flaws in OT processing of limit orders does not happen to run rampant, then the comparisons would be apples to apples. Thus far I have not seen that (unless I'm mistaken in what methodology has been used).

Conclusion - I believe that Limit orders based on prior Close will, over time, provide superior profits and lower drawdowns (did not discuss it but same logic applies) for short term RTM trades, IF the account is kept invested to a comparable degree.

Thread moved by JimDean on 8/4/2014 8:56 AM from Markets & Methods > OmniVest, Money Mgmt & Risk Control > Limit Orders versus MOO/MOC

Top of the page Bottom of the page
JimDean
Posted 7/7/2013 4:01 PM (#4988 - in reply to #4973)
Subject: Limit Orders versus MOO/MOC



Owner/Admin

Posts: 3925
2000100050010010010010025
Location:
USA: GA, Lawrenceville
After my last post, some additional suggestions and PortSim runs related to MOC orders were submitted in the OVest forum thread mentioned above. It's normally true, for a fast CPU and single-user system that OPilot-automated analysis + order submission can be done from a local machine without too much "lead time" required before the bell. Not so, for OVest architecture. Here are my comments about that:

---

There is an important fact to consider.

The servers must run their massive set of calculations PRIOR TO the close of the day, in order to place a MOC order at or before the close of each day.

Nirvana has already far surpassed my expectations in server response time but as the number of users continues to increase, it would seem logical that the time to do the computations also will increase. So lets break this down and think about it pragmatically and practically:

1. In addition to the first 15-30 or so min of a typical trading day, the time of highest trading activity is usually the last 15-30 min of the day.

2. Nirvana has not told us how long it takes to do the server calc's, but the emails seem to get sent out between 1-2 hrs after the close usually. Some of that delay has to do with email processing time and some has to do with what order they are sent in. There is presumably a comparable amount of info-distribution time required for OV-to-TP communications. So, barring a massive and expensive increase in server CPU cycles and bandwidth, my guess is that it's safe to assume there is a minimum of a one hour delay from the time that the signal processing number crunching starts, and the time that the TP's can get hold of the data - assuming no glitches.

3. The TP itself needs very little time to submit the orders to the brokers, but there is for some brokers a time limit for when MOC orders must be submitted.

4. There is zero time available in this scenario for users to review the signals if they choose to prior to TP submission - this is not an essential step due to the nature of how OVest works but its part of a prudent traders process so I'm listing it here as well.

Putting all that together for a BEST CASE SCENARIO, I would deduce that the price data upon which the MOC signals would be based would be at least 30m before the close, and more likely 45-60m.

Keeping point #1 in mind, I think it's safe to say that conditions one hour before the close is NOT equivalent to the conditions at the Close, in a similar nature as to the discrepancy between Close and next Open. Sure, there can be big gaps overnight but they are not that common for most stocks, statistically.

Therefore the conclusion that I come to is that EITHER approach, due to the delays, is going to be "winging it" to some degree. It's not possible currently with OT to BackTest/simulate the one hour early calc, btw, though it is possible with OPilot to trade that way at the HRE. So, any comparisons between MOC orders vs next day MOO orders are going to have 30-60m's worth of "slippage" in the modelling.

When OT+OPilot are run in a small list with one-two strats this one-hour "slippage" delay is not an issue. Kick off the run maybe 10m ahead of time on a fast machine and the MOC orders (or Market orders) can be placed before the closing bell. But, afaict, the server horsepower in combo with the large and growing number of strats in combo with the growing OVest customer base will always force a significant lag.

So, for the near-mid future, I believe that Limit orders the next day based on close of prior day, PLUS some improvements in OV-TP communications to make sure that all dollars are in play effectively regardless if the partial/no fills from some Limit orders, is the best way to go for very-short term strats like RTM.

Ideally the TP can be made smart enough to submit limit orders based on the "top tier" of allocated opening positions, then after a user-input delay (maybe 5-15 min?), can go back and evaluate what got filled and what didn't, and then RE-allocate the unused funds IMMEDIATELY (within the first 30m) to other Limit-orders for alternative trades that didn't make the first cut. That way those unused funds would not have to wait till the next day's open to be put into play.

Thread moved by JimDean on 8/4/2014 8:56 AM from Markets & Methods > OmniVest, Money Mgmt & Risk Control > Limit Orders versus MOO/MOC

Top of the page Bottom of the page
Jump to page : 1
Now viewing page 1 [50 msgs/pg]
( E-mail a Link | Printer Version | Search Room )

Owner of site: Jim Dean -- Forum content is confidential, and may not be distributed without written permission.