Price movements in the same direction?

kkseal

Well-Known Member
#32
Here's another condition that might be worth testing :-

If prices have gapped up or down yesterday, then they are likely to continue in the same direction today.

The conditions are

1) Gap Up
If yesterday's low > day before yesterday's high then if today's close > yesterday's close, plot 1 else plot 0

2) Gap Down
If yesterday's high < day before yesterday's low then if today's close < yesterday's close, plot 1 else plot 0

Regards,
Kalyan.
 

oxusmorouz

Well-Known Member
#33
Here's another condition that might be worth testing :-

If prices have gapped up or down yesterday, then they are likely to continue in the same direction today.

The conditions are

1) Gap Up
If yesterday's low > day before yesterday's high then if today's close > yesterday's close, plot 1 else plot 0

2) Gap Down
If yesterday's high < day before yesterday's low then if today's close < yesterday's close, plot 1 else plot 0

Regards,
Kalyan.
Yes,Kalyan. Will do that.
 

beginner_av

Well-Known Member
#34
good! very good indeed! btw what are you testing it on? each instrument has its own personality (CV of course!). What are your in sample and out of sample results? Are you trying out of sample walk-forward testing. if you use TS, theres a Grail genetic optimizer (saw it on some chinese board) . I am sure CV will have a very good understanding of it. May be he can help.
 

oxusmorouz

Well-Known Member
#35
Case 2 review : The relationship that exists between open and close on day 1 is not a significant indication for the next day's relationship.

Case III:
Objective: To determine if gap up or gap down affects next day's closing price movement in the same direction.

Data used:
Dow Jones Industrial Average from 1970 - Daily data. Total bars = 8,241 bars of data.

Condition: (Borrowed from kkeal)
The conditions are

1) Gap Up
If yesterday's low > day before yesterday's high then if today's close > yesterday's close, plot 1 else plot 0

2) Gap Down
If yesterday's high < day before yesterday's low then if today's close < yesterday's close, plot 1 else plot 0

The sum of the above is added and divided by the total gaps that have occured within the sample space.

Probable inferences:
a: If the probability is significantly higher than 50%, then,today's gap can be used as a tool for tomorrow's price in the same direction in which the gap has occured.

b: If the probability is significantly lower than 50%, then, today's gap can be used as a tool for tomorrow's price in the opposite direction in which the gap occured today.

c: If the calculated probability is neither significantly higher than 50% or significantly lower than 50%, indifference exists.

I use 50% mark here because that is the probability through coin toss.

Inference:
From 1970-2002, there occured a total of 112 gaps, out of which 51 were gap ups and 61 were gap downs.

For Gap Ups: 60.7843% of the times, the closing price had moved up from the closing level during the day in which the bar has occured.The remaining 39.2157% of the time, the closing price had moved down.

For Gap Downs: 57.3770% of the times, the closing price had moved down from the level in which the gap has occured. .The remaining 42.6230% of the time, the closing price had moved up

Overall: 58.9286% of the times, the closing price had moved in the direction of the gap.
Though this is an improvement from the previous cases, we must note that the data involved is low (just 112 gaps in the 8000bars of data).

I have attached the test results.
I request members to comment on this.

Obliged.
 
Last edited:

oxusmorouz

Well-Known Member
#37
good! very good indeed! btw what are you testing it on? each instrument has its own personality (CV of course!). What are your in sample and out of sample results? Are you trying out of sample walk-forward testing. if you use TS, theres a Grail genetic optimizer (saw it on some chinese board) . I am sure CV will have a very good understanding of it. May be he can help.
Till now, we have used only Dow Jones Industrial Average for testing purposes. Walk forward test would be good but data constraint is a killer :)
For eg, in case III, there were only some 112 bars in 35years of data (don't know if the data is flawed but I doubt it). Walk forward test would worsen the case by giving extreme results because of the small sample size.

if you use TS, theres a Grail genetic optimizer (saw it on some chinese board) . I am sure CV will have a very good understanding of it. May be he can help.
No Av unfortunately, I use metastock...find it comfortable... :(
 

SavantGarde

Well-Known Member
#38
I don't know if it applies to this thread, but probably worth a look...

In Bollinger Band Study in conjunction with Candle Stick:

For Bollinger Band (Settings: Source - Close, Period - 9, Std.Deviation - 2, MA - Simple)

Condition For A Compelling Buy the next day would be a close below the Lower Bollinger Band.

Example as of today would be BAJAJAUTO which has sort of closed almost at the Lower BB, Let us see what happens on Monday.

Other Possibilities with this study being that it may close for another few sessions below the lower BB, but eventually does move up quite substantially.

Do pardon me if this doesn't belong here, but thought of sharing it.

Happy & Safe Trading

SavantGarde
 

oxusmorouz

Well-Known Member
#39
I don't know if it applies to this thread, but probably worth a look...

In Bollinger Band Study in conjunction with Candle Stick:

For Bollinger Band (Settings: Source - Close, Period - 9, Std.Deviation - 2, MA - Simple)

Condition For A Compelling Buy the next day would be a close below the Lower Bollinger Band.

Example as of today would be BAJAJAUTO which has sort of closed almost at the Lower BB, Let us see what happens on Monday.

Other Possibilities with this study being that it may close for another few sessions below the lower BB, but eventually does move up quite substantially.

Do pardon me if this doesn't belong here, but thought of sharing it.

Happy & Safe Trading

SavantGarde
Thank you, SavantGarde :)
I made use of the metastock explorer and used the last 2500 bars of daily data on 30sensex stocks, sensex and DJIA.
I have attached the historical results for the same.

Obliged.

(metastock code for the same:
a:= BBandBot(C,9,S,2); {Bollinger Band details - Close,9 day SMA, 2 StdDev}
b:= L < a; {Basic Condition}
d:= BarsSince(b) = 1;
f:= C > Ref(C,-1) AND d; {Required condition}

100*Cum(f)/Cum(d))
 
Last edited:

Niranjanam

Well-Known Member
#40
Hello Friends

I think we are on the right track.There is substantial evidence that indicates stocks that perform the best (worst) over a three to 12 month period tend to continue to perform well (poorly) over the subsequent three to 12 months. Momentum trading strategies that exploit this phenomenon have been consistently profitable in the United States and in most developed
markets. Similarly, stocks with high earnings momentum outperform stocks with low earnings momentum.
To make money consistently , the trading methods should be based on statistically proven central tendencies of the market. Momentum based systems will work in all markets.

Niranjanam
 
Last edited: