|
PureBytes Links
Trading Reference Links
|
This is a function (with a signal) for Montecarlo's process.
It should be work, but it's freezes my TS2000...could an EL programmer of
the
list to verify if it's good or not (and modify it if it's necessary)?
Mario
{
User function: MonteCarlo
Monte Carlo simulation for position sizing.
This function records the profits and losses from the trading system that
calls it
and generates random sequences of the trades. For each sequence, it
calculates the return and drawdown assuming a fixed percent of the account
is
risked on each trade. The result is the probability that with a given
account size and risk percentage you can meet given rate of return and
drawdown
goals. For example, it might calculate the probability that you can
achieve
a rate of return of 50% with 30% drawdown by risking 10% on each trade.
ATTENTION TRADESTATION 4.0 USERS
Because this function calls the built-in function Random, which is new to
TradeStation 2000i, this function will not work in TradeStation 4.0.
I have developed a version of MonteCarlo with additional functionality and
several input and output options that will work with either TradeStation
2000i
or 4.0. This version can be purchased on my web site.
Copyright 2001 Breakout Future
www.BreakoutFutures.com
mrb@xxxxxxxxxxxxxxxxxxx
}
Input: ASize (NumericSimple), { account size, $ }
RetGoal (NumericSimple), { rate of return goal, %}
DDGoal (NumericSimple), { max closed out trade drawdown, % }
RiskPer (NumericSimple), { percentage risk per trade }
TrRisk (NumericSeries), { risk for current trade, $ }
NRand (NumericSimple); { number of random sequences }
Array: Trades[2000](0), { Trades, $ (win +; loss -) }
TrRisks[2000](0), { trade risks, $ (+) }
RandNums[2000](0), { array of random numbers }
RandIndx[2000](-1); { randomly chosen indices for trade array }
Var: NTrades (0), { Number of trades }
iSeq (0), { sequence number }
MaxNum (0), { max number left in random number array
RandNums }
iMax (0), { index (location) of MaxNum in array RandNums }
NCon (0), { number of contracts }
Equity (0), { account equity }
NewEquity (0), { account equity }
EqtyHigh (0), { highest equity }
DD (0), { closed trade drawdown }
DDmax (0), { worst case closed trade drawdown }
PReturn (0), { percent return }
AveRet (0), { average percent return }
ProbRet (0), { probability of return goal }
ProbDD (0), { probability of drawdown goal }
ProbComb(0), { probability of return & DD goals }
ii (0), { loop counter }
jj (0), { loop counter }
k (0); { loop counter }
{ Collect profit/loss and risk }
If TotalTrades > NTrades then Begin
If TotalTrades - NTrades < 2 then Begin
NTrades = NTrades + 1;
Trades[NTrades - 1] = PositionProfit(1);
TrRisks[NTrades - 1] = TrRisk;
end
else Begin { case where TotalTrades increments by 2 from one bar to
next }
NTrades = NTrades + 1;
Trades[NTrades - 1] = PositionProfit(2);
TrRisks[NTrades - 1] = TrRisk[1];
NTrades = NTrades + 1;
Trades[NTrades - 1] = PositionProfit(1);
TrRisks[NTrades - 1] = TrRisk;
end;
end;
{ Calculate results 1 day from last bar on chart }
If Date = JulianToDate(DateToJulian(LastCalcDate) - 1) {LastBarOnChart}
then Begin
for iSeq = 1 to NRand Begin { loop over # of random
sequences }
{ Generate random sequence of trades }
For ii = 0 to NTrades - 1 Begin { generate random numbers }
RandNums[ii] = IntPortion(NTrades * Random(10));
end;
For ii = 0 to NTrades - 1 Begin { sort index of RandNums
array }
MaxNum = -1;
For jj = 0 to NTrades - 1 Begin { find biggest remaining
number }
If RandNums[jj] > MaxNum then Begin
MaxNum = RandNums[jj];
iMax = jj;
end;
end;
RandIndx[ii] = iMax; { record location of max
number }
RandNums[iMax] = -1; { "remove" max number from
array }
end;
{ Calculate account balance and drawdown for current sequence of
trades }
Equity = ASize;
EqtyHigh = Equity;
DDmax = 0;
For ii = 0 to NTrades - 1 Begin
NCon = IntPortion(RiskPer/100 * Equity/TrRisks[RandIndx[ii]]);
NewEquity = Equity + NCon * Trades[RandIndx[ii]];
{ Calculate closed trade percent drawdown }
If (NewEquity > EqtyHigh) then
EqtyHigh = NewEquity
else Begin
DD = 100 * (EqtyHigh - NewEquity)/EqtyHigh;
if (DD > DDmax) then
DDmax = DD;
end;
Equity = NewEquity;
end; { for ii }
PReturn = 100 * (Equity - ASize)/ASize;
{ Accumulate results for probability calculations }
AveRet = AveRet + PReturn;
if (PReturn >= RetGoal) then
ProbRet = ProbRet + 1;
if (DDmax <= DDGoal) then
ProbDD = ProbDD + 1;
if (PReturn >= RetGoal and DDmax <= DDGoal) then
ProbComb = ProbComb + 1;
End; { for iSeq }
{ Calculate probabilities by dividing sums by number of sequences }
AveRet = AveRet/NRand;
ProbRet = 100 * ProbRet/NRand;
ProbDD = 100 * ProbDD/NRand;
ProbComb = 100 * ProbComb/NRand;
MessageLog("RiskPer=", RiskPer:5:1," RetGoal=", RetGoal:5:1,"
DDGoal=", DDGoal:5:1,
" ProbRet=",ProbRet:5:1," ProbDD=",ProbDD:5:1," ProbComb=",
ProbComb:5:1);
MessageLog("Average Return = ", AveRet:6:2);
end;
MonteCarlo = 1;
{
EasyEmini
A simple short-term e-mini breakout system to illustrate the use of
Monte Carlo simulation for position sizing.
System: Buy at a fraction of yesterday's range from yesterday's high. Hold
1 day then look to exit on the first profitable open. Use a fixed money
management stop (nominally $1000 in the e-mini).
WARNING: This system looks ahead one day to the next day's open to decide
whether
to exit the open position. Consequently, the system never executes on the
last bar
of data on the chart. This means that the entry order for the next day
will not
appear in the TradeStation Tracking Center.
ATTENTION TRADESTATION 2000i USERS
This system includes a call to function MonteCarlo, which performs Monte
Carlo
simulation for position sizing. MonteCarlo only works in TradeStation
2000i
because it uses built-in function Random, which is new to 2000i. I have
commented out the call to MonteCarlo so that this system will work
properly for
TradeStation 4.0 users. If you are a TradeStation 2000i user, you can
download
function MonteCarlo for free from my web site (address below). Remove the
comment braces {..} from the call to the line below that begins "Value1 =
...".
This will enable the call to MonteCarlo.
The MonteCarlo code records the profits and losses from the trading
system and
generates random sequences of the trades. For each sequence, it
calculates the return and drawdown assuming a fixed percent of the account
is
risked on each trade. The output is the probability that with a given
account size and risk percentage you can meet given rate of return and
drawdown
goals. For example, it might calculate the probability that you can
achieve
a rate of return of 50% with 30% drawdown by risking 10% on each trade.
I have developed a version of MonteCarlo with additional functionality
and
several input and output options that will work with either TradeStation
2000i
or 4.0. This version can be purchased on my web site.
Copyright 2001 Breakout Future
www.BreakoutFutures.com
mrb@xxxxxxxxxxxxxxxxxxx
Input: ASize (30000), { account size, $ }
RetGoal (50), { rate of return goal, %}
DDGoal (30), { max closed out trade drawdown, % }
RiskPer (10), { percentage risk per trade }
EntFrac (0.75), { multiple of prior day's range for entering }
MMStop (1000), { money management stop, $ }
NRand (1000); { number of random sequences }
Var: EntPr (0), { Entry target price }
XitPr (0), { mm stop exit price }
TrRisk (0); { trade risk, $ }
TrRisk = MMStop;
{ Entry conditions }
EntPr = H + EntFrac * (H - L);
If C > C[1] then
Buy next bar at EntPr Stop;
{
{ Exit conditions }
Exitlong("MMStop") next bar at EntryPrice - MMStop/BigPointValue stop;
If BarsSinceEntry >= 1 and open of next bar > EntryPrice then
ExitLong("ProfOpen") next bar at market;}
{ Perform Monte Carlo analysis: MonteCarlo only works in TradeStation 2000i
Value1 = MonteCarlo(ASize, RetGoal, DDGoal, RiskPer, TrRisk, NRand); }
}
----- Original Message -----
From: "Clive" <clivesmail@xxxxxxxxxxx>
To: "DH" <catapult@xxxxxxxxxxxxxxxxxx>
Cc: "Omega List" <omega-list@xxxxxxxxxx>
Sent: Saturday, April 20, 2002 2:25 PM
Subject: Re: AW: Monte Carlo Simulations
> Dennis
>
> I am starting to look into this and would appreciate if you could give
some
> more explicit details as how you do the simulation. Is it an add-on to TS
or
> a plugin for excel
>
> Thanks
>
>
> ----- Original Message -----
> From: "DH" <catapult@xxxxxxxxxxxxxxxxxx>
> To: "Omega List" <omega-list@xxxxxxxxxx>
> Sent: 20 April 2002 13:02 PM
> Subject: Re: AW: Monte Carlo Simulations
>
>
> | > What do you think is the right way of doing MC?
> |
> | I can't really comment on the portfolio approach. That gets complicated.
> | I trade index futures and I use the MC individually on each system. But
> | anyway, here's the general idea of how it works.
> |
> | First, you need to generate a list of profits and losses for the
> | backtest. Each backtested trade should risk an equal dollar amount.
> | That's the tricky part because the "risk" is defined differently in
> | different systems. Generally speaking, it's the maximum amount you can
> | lose on the trade. I usually use a large disaster stop based on ATR so
> | it's easy in that case.
> |
> | risk/trade = contracts * disaster stop
> | risk/contract = X * ATR
> | contracts (shares) to trade = Y * account size / ATR
> | solve for Y
> |
> | Once you have your list of trades (each trade adjusted for equal dollar
> | risk) you calculate the max drawdown. Store that number, shuffle the
> | trades into a random order, calculate the MaxDD again, and store that
> | number. Repeat a whole bunch of times - 10,000 is good. After you have
> | the list of 10K drawdowns, sort them in order and play with the numbers
> | statistically.
> |
> | In the end, you should generate a table like the one below. This is for
> | a real system. Every system will be different. Notice how trading just a
> | little bit smaller can dramatically reduce the risk of blowing out the
> | account with this particular system. I consider that valuable
> | information.
> |
> | As an aside, using Vince's methods, I get an Optimal_F of 26% which is
> | consistent with the MC results. The MC says, if I risk 26% on each
> | trade, there is a 99% chance of going broke. Vince fans will say you
> | should find Opt_F and then back off the leverage some until you feel
> | comfortable. The problem is they don't give you any way to calculate
> | "some." The MC helps you decide how much of the account to risk to stay
> | within your own particular comfort level.
> |
> | Probability of ruin (drawdown bigger than account size)
> | Percent of account risked on each trade
> | 0.01% 5.3%
> | 0.1% 6.1%
> | 0.2% 6.7%
> | 0.3% 7.0%
> | 0.4% 7.2%
> | 0.5% 7.4%
> | 0.6% 7.5%
> | 0.7% 7.7%
> | 0.8% 7.8%
> | 0.9% 7.9%
> | 1% 7.9%
> | 2% 8.6%
> | 3% 9.1%
> | 4% 9.5%
> | 5% 9.7%
> | 6% 10.0%
> | 7% 10.2%
> | 8% 10.4%
> | 9% 10.6%
> | 10% 10.8%
> | 11% 11.0%
> | 12% 11.1%
> | 13% 11.2%
> | 14% 11.4%
> | 15% 11.5%
> | 16% 11.7%
> | 17% 11.8%
> | 18% 11.9%
> | 19% 12.1%
> | 20% 12.2%
> | 25% 12.8%
> | 30% 13.3%
> | 35% 13.8%
> | 40% 14.3%
> | 45% 14.8%
> | 50% 15.3%
> | 55% 15.8%
> | 60% 16.3%
> | 65% 16.8%
> | 70% 17.4%
> | 75% 18.0%
> | 80% 18.7%
> | 85% 19.6%
> | 90% 20.6%
> | 95% 22.3%
> | 99% 25.4%
> | 99.9% 28.8%
> | 99.99% 32.5%
> |
> | --
> | Dennis
> |
> |
>
>
|