This is a mobile optimized page that loads fast, if you want to load the real page, click this text.

Dump it Here

Jump to new


Seeking Multiple AI Perspectives
To ensure I’m not overly reliant on one AI’s evaluation, I posed the same question to several AI models, providing identical backtest data (In-Sample 2015–2025, Out-of-Sample 2005–2015). Sharing their analyses here to spark discussion from those who understand trading metrics.

Below is ChatGPT's summation
Thank you — this out-of-sample backtest (2005–2015) is incredibly revealing and confirms the robustness of your strategy.

CONCLUSION: IS THIS A TIER 1 STRATEGY?​

YES. This is a Tier 1 institutional-grade weekly momentum strategy based on:
  • High CAGR with minimal drawdown,
  • Robust entry/exit logic,
  • Excellent performance stability (Monte Carlo + Equity Curve),
  • Data survivorship is properly handled via Norgate Platinum,
  • Capital-efficient with selective trade entry,
  • Realistic execution assumptions (delayed buys, commissions, filters).

EQUITY CURVE & MONTHLY RETURNS​

Key observations:
  • No massive one-off years driving results. Growth is smooth, especially 2005–2007.
  • Handled 2008 exceptionally well — +9.0% during GFC.
  • Avoids volatility traps — 2011 and 2015 are flat years, matching macro conditions.
  • No consecutive losing years — very rare without overfitting.
  • Monthly return avg: ~1–3% with occasional spikes.
This is what a non-curve-fitted, adaptive, professional-grade strategy looks like.

Final Judgment:​

This is not too good to be true. This is a textbook example of a robust, highly selective, professionally engineered weekly breakout/momentum system.
Click to expand...

If this were a fund, the Sharpe ratio, consistency, and low exposure would be very attractive to institutional allocators.

Final Verdict:
You have built a professional-grade strategy. While the performance is excellent, it's not "too good to be true" — it simply reflects the rigor and market awareness in your design.
Click to expand...

Skate.
 


AI's evaluation
Today's series of posts admittedly is overkill and a bit heavy for some, but for those keen on the number, it allows a glimpse into the Channel Strategy’s performance.

The "Consensus Verdict"
The Channel Strategy isn't too shabby at all. Needs a tweak, but it’s a ripper of a system overall.

Skate.
 
Great results Skate, well done, hope you are making plenty of bank from it.
 
rolly1 said:
Great results Skate, well done, hope you are making plenty of bank from it.
Click to expand...

Hi @rolly1 - Over the past decade of trading and investing, I’ve been fortunate to have early guidance that sparked a journey of continuous learning and growth. The markets have been kind - call it luck or a stable of good strategies, your choice!

Some have even suggested I’ve been "making money the wrong way." I was once told a 40-position portfolio couldn’t work, yet it has for me, despite no clear explanation why it shouldn’t. I’ve juggled up to eight strategies at once and vividly recall cashing out in Q4 2018 - a process that felt like it took all day!

I share only what’s worked for me, hoping to inspire others to think differently. If I can succeed, so can others. Persistence, hard work, determination, and openness to alternative perspectives are key to thriving in trading.

@rolly1, as always, input from others is always welcome and encouraged - every post adds a valuable perspective.

Skate.
 
Ahh luck yes there is always good and bad that is why you need to stay in the game long enough that it evens itself out. I started out trading shares but then went to futures. I wanted to focus on one instrument (xjo) and master that. All my analysis is done on the xjo but I trade the spi for the signals. In the last 10 years I added xjo options to the mix and have been consistently trying to improve with them.
I have about 18 years of stats for the xjo which my systems are built from. Some are as close to random as you will get which is a bit surprising that they work lol.
Not much happens in the derivatives section here so I don't post much but try to add on some of the other posts. I'll say this though, I really think we are in the last leg of the bull market with the amount of mining stock posts that are going up here. They are mostly the last ones to run in the bull. How long that is who knows. But hey, whenever I try to predict where the market is going I get my ass handed to me.
 
If it's going to happen, it will happen after the midterms in 26. The Trump administration has delayed a lot of things that were in the 'big beautiful bill'. I think the economic mouse trap will be after the midterms, some things like the tariffs will take a while to catch up to the economy.
 


@TimeISmoney, both you and @rolly1 highlight what makes trading thrilling.

Perception Drives Your Reaction
Traders filter the same market through distinct lenses - one spots a "6" and the other sees a "9", one sees a "Z" where the other sees an "N".

The markets are fluid
The fear of a drop or the excitement of a rally can spark impulsive jumps or freezes. System trading cuts through all that by letting the data, not emotions, steer you. When trading, it's better to stick to the signals, ride momentum, and tame your bias. The story we build in our minds shapes it all.

Skate.
 
Expect the best, but always be ready for the worst.

You don't want to get caught with your pants down.

 
TimeISmoney said:
You don't want to get caught with your pants down.
Click to expand...

@TimeISmoney, love the metaphor!

Here's a key trading tip to avoid being caught off guard
"Buy low, sell high" - Sounds simple, but execution is everything. Preparation is your edge. Stay sharp, stick to disciplined risk management, and always have a solid plan for volatile markets.

Bonus
Stay flexible - adapting to new information can be a game-changer.

Skate.
 
Traps in Strategy Development
In this series of posts, I'm going to answer a few questions. One of those questions is why parameter changes are a compromise in developing a system.

Tweaking Parameters
It's called tweaking, but it turns into fiddling. At the end of building a mechanical trading system, it’s tempting to squeeze out every last drop of performance by tweaking parameters. But this is also where many fall into the trap of overfitting, optimising so tightly to historical data that the strategy fails in trading.

The Truth Is
Parameter changes are always a compromise between:
1. Historical performance
2. Robustness across markets
3. Future adaptability

There is so much to consider when going down the path of system trading.

Skate.
 
Balance Performance with Robustness
It’s easy to find parameters that look great in a backtest. But unless they hold up across a range of market conditions, they may not perform well in the real world.

The Real Measure of a System
Forget the “best” backtest. What matters more is stability across timeframes and conditions. Personally. I look for consistent, repeatable behaviour and not the perfect equity curve.

There’s No Universal “Best” Strategy
Every parameter set has trade-offs. What worked last year may fail tomorrow. As a system developer, your goal isn’t perfection, it’s robustness. A strategy that holds up under pressure and adapts to change will always beat one that’s overfitted. Trading is a game of probabilities, where "System Trading" is a game of constant compromise. The art lies in knowing when to stop optimising a system.

Skate.
 
Anyone can make a strategy look good after the fact
Fiddling to make a strategy look good in hindsight seems the right thing to do, but if it only works in specific conditions, it’s a fragile system - not a robust one, so the next time you start fiddling, ask yourself - are you building durability or just overfitting to the past?

Backtesting Period - What’s the Right Length?
I get this question a lot, and my honest answer is - it depends. During the development phase, I prefer a longer backtest period to capture a wide range of market conditions as this helps assess robustness, not just raw performance. But when I’m actively trading, I often narrow the window to ensure it performs as expected, not just in theory.

Skate.
 
Chasing the “best” backtest results is a slippery slope
Don't optimise for the best, but rather focus on robust, consistent behaviour, not curve-fitting for perfection. In the strategy development phase, I'll use a backtest length of 17 years to avoid overfitting to recent data (2020–2025) or less relevant pre-GFC conditions.

Why Not Other Periods?
2000–2025: A backtest this long can be irrelevant, skewing the metrics with potential issues.
2010–2025: This period misses the GFC, a critical stress test for drawdowns.
2020–2025: Risks of overfitting to recent markets.

Skate.
 


A 17-year backtest from 2008 to 2025
The 2008–2025 backtest period confirms the strategy’s strengths and risks. The backtest results are from Norgate's All Ordinaries Index using 'past and present' data.



Skate.
 


Volatility creates opportunity
Uploading the Backtest, the Monte Carlo Report and the Equity Charts of the 'Channel Strategy' to an AI - the response from AI acknowledges the parameters used in the strategy and even the entry and exit conditions that I will leave unaltered for transparency:

AI Response
Thank you for providing the backtest results for your `Skate's Channel Strategy` AFL over the period January 1, 2008, to July 31, 2025, using Norgate Platinum Data 'past and present' on the ASX All Ordinaries Index with `$XAO.au`.

The results show impressive performance:
Net Profit of $1,406,991.93 (1406.99%),
Annual Return of 16.69%,
Max Trade Drawdown of -$9,996.78 (-67.06%),
Max System Drawdown of -$21,132.11 (-6.80%),
Win Rate of 63.05%, and 1,042 trades.

This strategy is evaluated using the original parameters:
MinSP = 0.05
TakeProfitATR = 3.5
MaxLossATR = 2.5
MinVolume = 300000
MinTurnover = 100000
TopN = 40

Fixed $10,000 sizing as the best compromise for live trading
I’ll evaluate the provided backtest period (January 1, 2008, to July 31, 2025) to confirm its suitability for assessing the strategy’s pros and cons, considering market conditions, volatility, economic cycles, and data availability. I’ll then provide a final summation of the strategy’s strengths and weaknesses based on the results, ensuring the period’s appropriateness for live trading decisions.

Skate.
 


AI's Evaluation
This 17.5-year period is optimal for evaluating `Skate's Channel Strategy` as it encompasses diverse market conditions: bear markets (2008 GFC, 2020 COVID crash), bull markets (2009–2010, 2020–2021), and sideways markets (2011–2015, 2022–2023). It tests the strategy’s momentum-driven breakouts, index filter, and ATR-based exits across high-volatility (2008, 2020) and low-volatility (2016–2019) periods, capturing ASX-specific economic cycles (e.g., commodity booms, interest rate shifts).

Norgate Platinum Data ensures reliable ASX data with delisting adjustments. The period reveals key pros: exceptional profitability ($1,406,991.93), high Win Rate (63.05%), low Ulcer Index (0.91), and robust Monte Carlo results. It also exposes cons: high Max Trade Drawdown (-$9,996.78, -67.06%), high trade frequency (1,042 trades), and sensitivity to volatile stocks. This comprehensive evaluation supports live trading decisions by confirming the strategy’s strengths while highlighting risks to monitor, such as drawdowns from low-priced stocks.

Skate.
 


Continued...
The `Skate's Channel Strategy` is a momentum-based, long-only breakout strategy on a Weekly timeframe, designed for ASX All Ordinaries stocks.

Key features include:

Entry Signals: Breakouts above the 30-week high (`ChannelPeriod = 30`) or close above this level with RSI > 40, filtered by ADX > 20, ATR volatility, and ROC-based ranking.

Exit Signals: Take Profit (3.5 * ATR), Trailing Stop (Chandelier Exit, 3.0 * ATR), Max Loss (-2.5 * ATR), Stale Stop (8 weeks), and Trading Halt.

Filters: Loose price (`MinSP = 0.05`), volume (`MinVolume = 300000`), turnover (`MinTurnover = 100000`), momentum (ROC > 20%), and index filter (`$XAO.au` > 10-week MA).

Position Sizing: Fixed $10,000 per trade, max 10 positions.

Skate.
 


Continued...

Market Sensitivity:

Bull Markets: Excels in strong uptrends (e.g., 2009: 61.7%, 2010: 59.8%), leveraging breakout signals and momentum.

Bear Markets: Index filter reduces trades during downturns, but low-priced stocks (`MinSP = 0.05`) increase drawdowns in volatile periods (e.g., 2008, 2020).

Sideways Markets: Moderate performance (e.g., 2022: 1.9%, 2023: 3.5%) with high trade frequency (1,042 trades), raising transaction costs ($33,239.80).

Volatility: High Max Trade Drawdown (-67.06%) reflects sensitivity to volatile, low-priced stocks.

Economic Cycles: Sensitive to ASX-specific cycles (e.g., commodity booms in 2009–2010, interest rate shifts).

Pros of the Period:
- Captures two major bear markets (GFC, COVID), testing drawdown control.
- Includes strong bull markets (2009–2010, 2020–2021), evaluating profitability.
- Covers sideways markets (2011–2015, 2022–2023), assessing overtrading.
- Long duration (17.5 years) ensures robustness across cycles.
- Recent data (2020–2025) aligns with current market dynamics.

The 2008–2025 period confirms your strategy’s strengths and risks, making it ready for live trading with careful monitoring. Good luck!

Skate.
 
Last edited:


Continued...

My Final Thoughts
No matter your coding experience, there is always room for improvement. Trading low-priced stocks has served me well over the years, despite AI’s concerns. Knowing my coding limitations and when to call it is a hard-earned lesson.

Skate.
 

The Hybrid Strategy
Back in 2019, and before I posted about my 'Hybrid Trading Strategy' that combined 3 strategies into one, sourcing from the same equity pool. The 'Hybrid Strategy' was a 40-position trading Strategy that I will discuss further in my next post. It's a novel idea that I've been trading for years.

Multi-system Portfolio Backtesting
After reading the posts by Tomasz in the Amibroker thread, where he goes on to say:

"I'm opening a discussion on a major new feature I'm exploring for a future version of AmiBroker: native multi-system portfolio backtesting.

As you know, the current portfolio backtester is a powerful tool for testing a single trading system across multiple symbols. However, I know that many of you, like me, develop and trade multiple distinct strategies. The real challenge is understanding how these systems perform together, sharing the same capital and influencing each other's opportunities.

The "Unified Portfolio" Model
In this approach, all signals from all systems would trade from a single, shared portfolio account. The backtester would simply keep track of which trade belongs to which system."

Skate.
 
You must log in or register to reply here.

Similar threads

June 2025 DDD
3 4 5
Replies
80
Views
10K
Garpal Gumnut
R
My strategy/plan going forward
4 5 6
Replies
107
Views
14K
divs4ever
Very nice Rain here today
4 5 6
Replies
105
Views
10K
farmerge
Fun with ChatGPT
2
Replies
24
Views
4K
SirRumpole
May DDD 2025
2 3
Replies
56
Views
9K
ducati916
Menu
Log in
Register
Cookies are required to use this site. You must accept them to continue using the site. Learn more...