-
The Forum works well on MOBILE devices without an app: Just go to: https://forum.tsptalk.com
-
Please read our AutoTracker policy on the IFT deadline and remaining active. Thanks!
div>
$ - Premium Service Content (Info) | AutoTracker Monthly Winners | Is Gmail et al, Blocking Our emails?
Find us on: Facebook & X | Posting Copyrighted Material
Join the TSP Talk AutoTracker: How to Get Started | Login | Main AutoTracker Page
The Forum works well on MOBILE devices without an app: Just go to: https://forum.tsptalk.com ...
Or you can now use TapaTalk again!userque & NUGT/DUST
- Thread starter userque
- Start date
userque
Well-known member
I can do either/both. That's one of the changes I made. I can do like before, like I indicated earlier with specific times. If you would also like a midday call, what time frame would you want me to strive for?
Also, an early call should only be based upon the opening price.If it is not too much trouble, I would prefer both. That is the call based on just the open (post whenever convenient) and the one based on open, high, and low. I like how you were doing the preliminary call about 20 minutes before the close and then updated it 5-10 minutes before the close.
userque
Well-known member
Will do.
BTW/FYI...I only update the preliminary call 5-10 min. before close if there is a change, otherwise, I "update" it about 1 min. before close just to delete the temporary language from the post.
userque
Well-known member
Unofficial, early forecast:
Based only on today's Open and possibly prior day's data and dates, the unofficial and preliminary forecast for today's close is to hold nugt.
This forecast may be overruled by the official forecast due later today.
_______________________________________________________________
DISCLAIMER: Userque's trading signals are not affiliated with TSP Talk or Buy Low Sell High, Inc. The information is for educational purposes only! The information is not advice or recommendations. The information may be revised at anytime. Userque does not give investment advice. Do not act on this data. Do not buy, sell or trade any stocks, ETF's, indices, funds, or any other instruments/vehicles based on this information. Userque may trade differently than discussed or posted in this forum.
userque
Well-known member
I have a buy set at 22 and change. If your holding I hope it doesn't get filled
LOL...fill it!...not my first rodeo
RealMoneyIssues
Well-known member
daaaaang...
userque
Well-known member
FOR 11-06-15 AT CLOSE:
ENTER DUST
_______________________________________________________________
DISCLAIMER: Userque's trading signals are not affiliated with TSP Talk or Buy Low Sell High, Inc. The information is for educational purposes only! The information is not advice or recommendations. The information may be revised at anytime. Userque does not give investment advice. Do not act on this data. Do not buy, sell or trade any stocks, ETF's, indices, funds, or any other instruments/vehicles based on this information. Userque may trade differently than discussed or posted in this forum.
Last edited:
userque
Well-known member
FAQ's
Your system wants to go into DUST after it has already been up 40%?
Recently...I think NUGT was already up 20 something %, then the system went into NUGT for a nice profit--also to the surprise of some. This is a trading question, not really an algorithm question. I have my plate full with doing and talking algorithms. I definitely don't have room on it to debate the feasibility of "buying high, and selling higher."
The system uses the past to predict the future. Past patterns. Its forecasts are based upon what has happened in the past; it doesn't "create" forecasts out of black-box formulas. At the most, it will average several past patterns into a forecast. If it didn't predict something correctly, then the current price action is mathematically outside of the norm, historically speaking--perhaps due to unexpected news (FOMC?), unpredictable factors, etc. So, historically speaking, this type of price action has happened in the past, and the system is merely predicting it to happen again.
Have you kept the previous versions so you can compare at all, or is this really due to "improvements?"
Some changes don't need testing because they simply enhance what the system already does. IOW The system will choose the "best" course of action; it will either be based upon the improvements, or the "original," which ever is "best." In still other words, such improvements can't make it worse.
Other changes I do test initially, to a statistical significance. But no, I don't continue testing and comparing each and every improvement, months/years into the future. That would require impossible amounts of time and computer resources, aside from being unnecessary.
The changes I've made are very recent...too soon to draw conclusions (especially in light of recent less predictable/unexpected news, i.e., FOMC news etc.), at least for me.
userque
Well-known member
FOR 11-09-15 AT OPEN:
ENTER NUGT
* SMARTER MODEL PREFERS TO BE LONG--EVEN IF GAP UP OPENING. SEE FOLLOWING POST
_______________________________________________________________
DISCLAIMER: Userque's trading signals are not affiliated with TSP Talk or Buy Low Sell High, Inc. The information is for educational purposes only! The information is not advice or recommendations. The information may be revised at anytime. Userque does not give investment advice. Do not act on this data. Do not buy, sell or trade any stocks, ETF's, indices, funds, or any other instruments/vehicles based on this information. Userque may trade differently than discussed or posted in this forum.
userque
Well-known member
Progress Update
Two Brains are Better Than One: OPEN Model
Decided to break up the problem. Instead of one GDX model, there are now two. One trades at/near the OPEN; the other at/near the CLOSE. In theory, splitting up the tasks should allow each algorithm to "specialize" and be more effective. I've tested the OPEN model. It did better than the previous model that traded only at the CLOSE. The test was based on the model buying at the OPEN, and selling at the CLOSE. (The Closing model will trade the close, if it hunts. The Closing model is designed to essentially, only trade the GAP.)
Wait for the Opening Before Deciding How to Trade the Opening
The OPEN model is working. In order to get the more recent price information, I designed it to use the current days opening price to forecast the direction going into the CLOSE. There are two ways to do this; I could simply wait for the opening, and then plug in the price, then make the trades if needed.
The other way is to create a "trading index." This index would list scores of possible opening prices for the upcoming trading day, alongside would be the model's forecast for each price. This can be generated well in advance and allows for very quickly knowing which trade to execute. It involves writing more code, but I hope to have it ready by Tuesday.
Tested and Rejected This Model
I also created and tested a model that forecasts a high price and a low price for the following trading day. Sell at the high, buy at the low. I tested it thoroughly, but it didn't perform as well as the others.
Discretionary Trades
Having a 'time' model could improve results (see below). Instead of buying at the opening, one may decide to wait until 10:30am for example. Even without such an edge, one could simple wait until the price stops falling at the OPEN before buying. If the price is rising, one would simply buy immediately. I will post what the system is doing each trade. Absent such a post, then the system is trading at the OPEN/CLOSE without my discretion.
CLOSE Model
I expect to have the Close Model ready by...the close. (Assuming it hunts. I will decide if it is better to trade the OPEN and the CLOSE, or just the OPEN. Not sure I'll be able to decide--conclusively--by the close.) It will give preliminary unofficial forecasts as before. No trade index will be generated as too many variables are involved (HIGH, LOW, LAST).
Future Models
I would like to test a model that forecasts buy and sell times using intraday historical data (I've got some laying around). But in the mean time, I'm keeping recent records of the times the lows and highs occur; to see if any basic statistical correlations or patterns exist. I would have to build a full model and use much more data for a deeper analysis.
In Closing
The models are evolving into their final form. Once that happens (soon), I can then shift to vastly (and slowly) increasing the size of their databases. That information will then let me know how to best choose the optimal parameters.
userque
Well-known member
Got a very late start on optimizing the new CLOSE Model. No early forecasts today. I'm not sure if there will be a forecast at all today for the CLOSE. If there is, it'll be more risky than usual with the short optimization period. So, it'll either be a last minute risky forecast; or a move to CASH at the CLOSE today.
userque
Well-known member
FOR 11-09-15 AT CLOSE:
HOLD NUGT
* CLOSE MODEL NOT READY. FORECAST MADE IN PART BASED UPON OLDER MODEL.
_______________________________________________________________
DISCLAIMER: Userque's trading signals are not affiliated with TSP Talk or Buy Low Sell High, Inc. The information is for educational purposes only! The information is not advice or recommendations. The information may be revised at anytime. Userque does not give investment advice. Do not act on this data. Do not buy, sell or trade any stocks, ETF's, indices, funds, or any other instruments/vehicles based on this information. Userque may trade differently than discussed or posted in this forum.
userque
Well-known member
FOR 11-10-15 AT OPEN:
HOLD NUGT
* 'WHAT IF' ANALYSIS MAKES WAITING FOR THE OPENING PRICE UNNECESSARY.
_______________________________________________________________
DISCLAIMER: Userque's trading signals are not affiliated with TSP Talk or Buy Low Sell High, Inc. The information is for educational purposes only! The information is not advice or recommendations. The information may be revised at anytime. Userque does not give investment advice. Do not act on this data. Do not buy, sell or trade any stocks, ETF's, indices, funds, or any other instruments/vehicles based on this information. Userque may trade differently than discussed or posted in this forum.
userque
Well-known member
FOR 11-10-15 AT CLOSE:
ENTER CASH
_______________________________________________________________
DISCLAIMER: Userque's trading signals are not affiliated with TSP Talk or Buy Low Sell High, Inc. The information is for educational purposes only! The information is not advice or recommendations. The information may be revised at anytime. Userque does not give investment advice. Do not act on this data. Do not buy, sell or trade any stocks, ETF's, indices, funds, or any other instruments/vehicles based on this information. Userque may trade differently than discussed or posted in this forum.
RealMoneyIssues
Well-known member
userque
Well-known member
Yeah. The CLOSE model doesn't do as well as the OPEN model. I posted about testing this, I think.
userque
Well-known member
FOR 11-11-15:
ENTER DUST AT OPEN IF GDX OPEN IS 13.87 OR LESS*
ENTER NUGT AT OPEN IF GDX OPEN IS 13.88 OR MORE*
ENTER CASH AT CLOSE
*IF THE GDX OPEN IS LESS THAN 12.47, OR GREATER THAN 14.47, THEN DISREGARD THE ABOVE. IN THIS CASE, I'LL POST THE REAL-TIME TRADE AFTER THE OPENING.
(The system ran a "what if" analysis testing what the forecast will be if tomorrow's GDX opening is any of multiple possible prices, starting from $1.00 less than today's close, up to $1.00 greater than today's close...incrementing the tested opening price one penny at a time. This should cover 99+% of the openings, in my guess. As show above, it switched for a DUST forecast to liking NUGT if the opening tomorrow is 13.88 or above--up to 14.47)
To be even clearer, the system didn't test for GDX openings below 12.47, or above 14.47.
_______________________________________________________________
DISCLAIMER: Userque's trading signals are not affiliated with TSP Talk or Buy Low Sell High, Inc. The information is for educational purposes only! The information is not advice or recommendations. The information may be revised at anytime. Userque does not give investment advice. Do not act on this data. Do not buy, sell or trade any stocks, ETF's, indices, funds, or any other instruments/vehicles based on this information. Userque may trade differently than discussed or posted in this forum.
Similar threads
