in only a few days i will close my first trade here at profit---i think so--
maybe i should have close this trade long ago to make new trades---nein nein---i will not have my first trade losing--i will not bring discredit to great manics board--kevin affords have great patience---no need to rush and fret--many profits lie in the future--this everyone will come to see--
herr shadow--i consider something new---maybe this will be added to my model--positive divergence off ma156---look at 10:35 bar this morning---many times i have seen this---i will go back in time and study this---please look at this for your opinion--
--now i watch and wait--
kevin--
der beobachter
Hello Kevin
ReplyDeleteI told you many months ago I liked your 156/234 "discovery", whether you should incorporate it into your model, I really can't say. It would have worked well Tuesday and yesterday. My guess is it would add risk. Maybe it would be worth the extra risk. You should reconsider your attitude about double-down and stop-loss. Not setting stops is a type of arrogance. Just my opinion.
Your EITHER/OR just signaled HOD at 57.66. It will be interesting to see if this holds.
ReplyDeleteOnce again this EITHER/OR signal was very accurate and very quick to signal. The 57.66 trade occurred at 1:44 and the signal came 7 minutes later when the price was still above 57.60. Though there was not much down movement off the high, still this signal keeps impressing me.
ReplyDeleteI don't know why I never thought of this before or why you never thought of it, but it is suddenly obvious to me that this signal needs to be modified so that instead of telling you the HOD or LOD just occured it needs to indicate the high of week or low of week just occurred. If not that, then the High of 2 day or 3 day. It should be possible to do this the same way we translated the 16-day middle line to the 5-minute chart. I will put in work hours with you if you want to work on this.
I am studying your EITHER/OR formula. It puzzles me a little why this thing works at all, let alone why it works so well.
ok--yes--i would like to try work on your idea--so far i see problems--
ReplyDeletei tried something on this problem this morning aber it is not good--
take the 'either/or' formula--do not change anything--just transfer it straight to 10min--then 15min--then 30min--then 60min--
ok--this is what you will see--up to 30min the formula still works fine---at 60 min starts to deteriorate--then next step up in time is to daily chart---formula is completely broke down at daily time frame--
ja--ja--ok--i already know what you will say---adjust the formula--trnslate the formula from 5min to daily time frame--
--i will tell you my thoughts why this will not work--
(one)--the 'either/or' does not always signal every single day on 5min--sometimes 2 days go by--no signal--
(two)--1 daily bar = 78 5min bars--
ok look at the equation backwards for translation from 5min to daily---you see we are dealing with a minimum period of 78 days--if 2 periods go by without a signal then we are dealing with a period of 234 days---
what good is a signal that signals once in 234 days?? maybe it is good in theory--not good for short term trading--this even assumes the signal would still be accurate after so much translation--how could i backtest such long periods?? problems problems problems--do you see my discouragement???
many hours of work would be in this problem---i start to see the law of diminishing returns--
if you have better thoughts tell me a new way to think about this problem--
this is a different subject---look how 156 and 234 come up again and again in every problem--in every timeframe--i wish i could understand this--
Hello Kevin:
ReplyDeleteI see what your saying. I want to think about this. Something tells me there MUST be some intermediate point between 5 minutes and 78 days. Maybe you are giving up too soon.
I have a bad flu and I'm not thinking clearly this morning. I can't find the spreadsheet on (EITHER/OR, 5-minute) backtesting you sent me last summer. If you could send me that in an e-mail attachment, I would like to look at it again in Excel.
How much trouble would it be to run those same backtests in the 30 minute timeframe?
If the 5-minute timeframe can signal the HOD/LOD then why couldn't the 30-minute time frame signal the 6-day HIGH or LOW? Maybe this is crazy. Just something to look at and think about.
I will watch my e-mail for the spreadsheet.
Thank you.
Hello Kevin,
ReplyDeleteI got the Excel attachment and the updated backtests. Thank you very much.
I think I've thought of a better way of approaching this now. I was wrong to think in terms of translating the formula into a longer time frame. This would be a waste of time, like trying to re-invent the wheel. And it would get into all the problems you mentioned in your post and in the e-mail. You already have an accurate signal for indicating HOD or LOD on the 5-minute chart. Let's just stick with that and build from that.
Let's look at it like this: We can expect the "EITHER/OR" to continue giving a fairly reliable HOD or LOD signal on the 5-minute chart. Eventually one of these signals, one of these HOD's (the last one) will not only be the HOD in the 5-minute chart, it will be the top of the upleg in the daily chart. We need to see if there is some way to determine that one of these signals is probably "the last one" at the time it is occurring.
So we need to go back and find the date of the tops of uplegs and bottom of downlegs in the daily chart. Then we need to see if there was an "Either/OR" signal that signaled on that day in the 5-min chart. Then we need to see if there was something different about these "Either/Or" signals that occurred at tops and bottoms and all the other "Either/Or" signals. The difference probably won't be in anything about the signal itself, it will probably be in the configuration of the charts (5-minute,daily) at the time the signal signals.
By configuration I mean the state of the MA crossovers and Price/MA crossovers, failure-in-oversold, etc. in the 5-minute, location of price in the cycle channel, relative RSI reading on the daily, etc. All the things we've looked at all along. Hopefully the backtesting won't get overwhelmed with too many variables. I'll split up the work with you. We can start out with just 4 or 5 variables and see if that points in any direction.
ok--yes--this seems right--very much--already today i am starting on this---
ReplyDeletei am defining tops and bottoms by the confirmation signal---
for defining position i am counting the last bars twice--so last bar of upleg is also first bar of downleg---last bar of downleg is also counted as first bar of upleg---there are no zero bars---just so we stay on the same page and there is no confusion--
i only have 5-min data back to sept 19---so this is really not enough---so maybe i will switch to a 30-min chart to go back farther---only if i must--
i have not chosen the variables--which do you believe are most important?? i belive i will start with 7 variables--i will tell you these before i begin---maybe you will want to add something--i am going to return to e-mail---i can attach charts and spreadsheets to e-mail--which is difficult on this blog board---when we finish you can post a chart here of what we learn---maybe nothing--i am hopeful and not discouraged--this plan seems possible--my hunches tell me this--
i will email every progress before friday--think about variables---this is important--
ja--this is great--already this shows promise---
ReplyDeletedec 19 downleg botom occurred @3:50--
---either/or signaled lod in 5min chart that day @3:47---either/or said lod 54.20---actual lod was 54.17---i can live with this--
now the hard part--distinguish this lod signal from all the lod signals that came before--if there is an answer kevin will find it--
i like this new project--
one more thing--i will only call an actual crossover a signal--if [ema20 = the band] this is no signal---it must actually cross over---
ema20 must read greater than the top band for lod---
ema20 must read less than the bottom band for hod--
equal does not count to say a signal occurred--