Current location | Thread information | |
![]() ![]() ![]() ![]() ![]() ![]() |
Last Activity 4/21/2021 9:48 AM 8 replies, 1597 viewings |
|
|
Printer friendly version |
^ Top | |||
SteveL![]() Veteran ![]() ![]() ![]() ![]() Posts: 262 Joined: 8/19/2005 Location: Boulder, CO ![]() |
As I've posted before, I've been using MoC entries and exits in simulation, but MKT orders near market close for actual trading. I currently have AT configured to start at 25 minutes prior to market close. Processing usually completes about 8 minutes before market close. Trade exits happen during processing, and AT places new entry orders at completion of processing. I'm running with an EOD profile. Therefore data is delayed 20 minutes. That means the data was last updated 45 minutes prior to market close due to the 20 minute delay in the OmniData EOD data feed. I thought it might be useful to show a comparison of simulated trades vs. actual trades for the past few weeks. On 8/31/2018, I added additional strategies to the ATM Macro & Micro profile that Mark Holstius shared, which also increased the number of signals. So, the comparison is only valid since that date. Prior to 8/31, you'll see the simulation shows more trades (due to more strategies in the simulation) vs actual trading. Since that date, there are similar number of trades. Where there are different trades (PortSim vs actual), the profit/loss profile is similar (eyeball analysis, not a rigid statistical analysis). This is a small sample set. But, it is consistent with what I've observed since I put ATM into use for trading near market close. In my opinion, trading with MKT orders near market close works very well. It enables exiting trades to free up funds for newly signaled trades. [Edited by SteveL on 9/23/2018 4:12 PM] ![]() | ||
^ Top | |||
Buffalo Bill![]() Legend ![]() ![]() Posts: 539 Joined: 10/3/2006 Location: Stafford, VA ![]() |
Thanks for the update. It makes sense, esp having to have your to-do list done by 3:45 EST (with a few minutes pad of course). This is something I hadn't considered: "It enables exiting trades to free up funds for newly signaled trades." Interesting. Any idea how many "extra" trades MKT (near close) exits allowed vs pure MoC? | ||
^ Top | |||
SteveL![]() Veteran ![]() ![]() ![]() ![]() Posts: 262 Joined: 8/19/2005 Location: Boulder, CO ![]() |
Buffalo, No, I don't know exactly, but I am frequently fully allocated. And, it will depend on your strategies and ATM method. I think you can answer your question by evaluating your trades. I am under the impression from a post you made that you are trading with MOC entry and exits. Does OT place both MOC entries and exits simultaneously and keep you fully allocated? From experiments I have done in the past, I came to the conclusion that didn't work in actual trading (NOTE: Simulation does allow simultaneous MOC exit/entry when "reuse intraday funds" is selected). So, anytime you are fully allocated and you have a trade exiting with a MOC order, were there any newly signaled trades that could have been taken if that exit had occurred prior to market close? [Edited by SteveL on 9/24/2018 12:00 AM] | ||
^ Top | |||
SteveL![]() Veteran ![]() ![]() ![]() ![]() Posts: 262 Joined: 8/19/2005 Location: Boulder, CO ![]() |
Another comment regarding MKT vs MoC orders... Occasionally in the past, OT AutoTrade/ATM processing has taken longer than expected, and orders were issued too late (market had closed). In that case, the late MKT orders are executed at market open - not my preferred entry time, but better than missing the trade with the RTM strategies. The longer than expected processing seemed to be caused by longer than normal data downloads. I have found that configuring AT to do it's "Update Portfolio" run 1 hr before market open seems to cure this problem (rather than doing the update in the afternoon/evening after market close. This gives me a more consistent AT run-time prior to market close. My guess is that historical data is updated overnight and that is what can sometimes cause a longer data download. (Pure speculation - pehaps Barry can comment.) | ||
^ Top | |||
Barry Cohen![]() Sage ![]() ![]() ![]() ![]() ![]() ![]() Posts: 6338 Joined: 1/19/2004 ![]() |
Nice analysis, Steve! Longer data downloads are rare, but are typical when we do a refresh of something like the MG groups. | ||
^ Top | |||
Jim Dean![]() Sage ![]() ![]() Posts: 3022 Joined: 9/21/2006 Location: L'ville, GA ![]() |
Barry:did you just refresh the symbols that make up the mg groups??? I thot that was kaput - that the symbols were based on the mg maps from like five years ago. If they’ve been refereshed, yaaay! You need to tell people! | ||
^ Top | |||
Buffalo Bill![]() Legend ![]() ![]() Posts: 539 Joined: 10/3/2006 Location: Stafford, VA ![]() |
Steve TBH I wasn't looking for it so I'm not sure! But, I like your idea because it eliminates the whole issue and allows some extra time for analysis (don't have to be done with to-do by 3:45 EST for IB MOC order cutoff) | ||
^ Top | |||
Barry Cohen![]() Sage ![]() ![]() ![]() ![]() ![]() ![]() Posts: 6338 Joined: 1/19/2004 ![]() |
Jim, I meant the MG chart data, not the constituents with the groups. Sorry to get your hopes up! | ||
^ Top | |||
Jim Dean![]() Sage ![]() ![]() Posts: 3022 Joined: 9/21/2006 Location: L'ville, GA ![]() |
Thanks for clarification. You were referring to the calculated MG averages. I hope that MG components updates can be put back into gear. They are a core feature of OT and VT that essentially have been “broken” a very long time. |
|
|
Legend | Action | Notification | |||
Administrator
Forum Moderator |
Registered User
Unregistered User |
![]() |
Toggle e-mail notification |