Justin's Playground

My Advantage Futures account has been recovered from the day trading loss I made months ago. I always recoup losses from day trading with profits from spread trading. I won't do this again. I will use only prop firm accounts for day trading.

Compared to 10 years ago, has modern day manual day trading become more difficult to make a consistent profit?


Yes — in many important ways, manual day trading today is generally more difficult to make a consistent profit than it was a decade ago. That doesn’t mean it’s impossible, but several structural changes in markets and technology have raised the bar for individual traders. Here’s why:
1. Markets Are More Efficient

1. Algorithmic and high-frequency trading (HFT) now dominate many markets. These systems exploit tiny inefficiencies in microseconds, which historically were opportunities for humans. Today, those opportunities are far rarer.
2. In markets like Forex, equities, and crypto, liquidity algorithms adapt so quickly that human reaction time can be a competitive disadvantage.

Result: Traditional patterns that once worked reliably are now arbitraged away faster.
2. Competition From Smarter Technology

1. Ten years ago, many discretionary traders still had an edge over slower institutional systems.
2. Now AI-assisted algorithms, machine learning models, and low-latency execution tools are widely used — even by some smaller funds.

Result: The playing field isn’t level; retail traders are competing with automated systems optimized for speed and pattern recognition.
3. Higher Costs & Tighter Spreads

1. Commissions have dropped, but spreads and execution costs can still eat into profits, especially in fast markets or low-volume stocks.
2. Some brokers add fees for order routing, data, or “smart” execution.

Result: More trades can mean more costs, making net profits harder to sustain.
4. Information Noise Has Exploded

1. A decade ago, news arrived via financial terminals and scheduled releases. Now:
   1. Markets react instantly to social media, Reddit forums, Twitter/X rumors, and macro surprises.
   2. Sentiment-driven spikes can overwhelm technical analysis signals.

Result: Noise has increased while signal clarity has decreased.
5. Retail Tools Have Improved — But So Has Risk

1. Professional-grade charting and backtesting are now accessible to retail traders.
2. But that also means crowded strategies: once a pattern becomes popular, it stops working as well.

Result: Tools alone don’t guarantee an edge if everyone uses the same ones.
6. New Asset Classes, New Dynamics

1. Crypto, micro-caps, and ETFs are popular day trading spaces — but each has distinct behavior:
   1. Crypto markets trade 24/7 with high volatility but unpredictable flows and often no circuit breakers.
   2. Micro-caps can be manipulated or illiquid.
   3. ETFs introduce arbitrage mechanisms impacting price movements.

Result: There’s opportunity, but it’s different and often harder to model consistently.
7. Knowledge Is More Widely Shared

1. Online trading communities, courses, and YouTube channels democratize strategies. That’s good — but when thousands use the same setups, edges get neutralized.

Result: Competitive advantage requires either unique insights or superior execution disciplines.
What Has Become Easier?
Yes, some aspects have improved:
AspectEasier Today?Why
Access to marketsLow-cost brokers, fractional shares
Real-time data & analyticsPowerful platforms and APIs
Learning resourcesOnline courses, mentorship
Execution toolsMixedGood tools available, but so are they to others
Overall Conclusion
Yes, manual day trading to make consistent profits is generally harder now than it was 10 years ago.
The main reasons:

1. Increased competition from automation
2. Fewer exploitable inefficiencies
3. Higher noise relative to signal
4. Costs and execution challenges

But it’s not impossible — profitability still comes from:

1. Strong discipline
2. Risk management
3. Unique strategies or niche edges
4. Adaptability, not old patterns

If you want, I can break down specific markets (e.g., stocks vs Forex vs crypto) or compare typical returns then vs now based on historical data and research.

Is Natural Gas calendar spread trading using seasonality still profitable?


Natural Gas calendar spread trading based on seasonality can still be profitable, but it’s not as easy or consistent as it was 10–20 years ago. Since you’re a day trader, I’ll frame this from both a swing and shorter-term spread perspective.
What We’re Talking About
Most traders mean trading spreads in:

1. Natural Gas (NG futures)
2. Exchange: New York Mercantile Exchange (NYMEX)

Typical seasonal spreads:

1. Winter vs Summer (e.g., Jan–Apr)
2. Shoulder months (Mar–Apr, Oct–Nov)
3. Storage-related spreads (Oct–Jan, etc.)

Why Seasonality Used to Work Better
Historically:

1. Predictable winter heating demand
2. Slower information flow
3. More physical hedging dominance
4. Fewer systematic/stat-arb funds

Seasonal storage cycles were more “mechanical.”
What Changed
1. LNG Export Growth
U.S. LNG exports made NG more globally sensitive.
Weather in Europe now impacts U.S. spreads.
2. Weather Models Are Faster
Everyone sees the same NOAA model updates instantly.
3. Algo & Spread Funds
Stat-arb desks trade NG spreads heavily.
Edges get arbitraged faster.
4. Production Volatility
Shale responsiveness dampens some traditional seasonal moves.
Is It Still Profitable?
Yes, but:

1. Pure “buy every March, sell every October” no longer works reliably.
2. Seasonality is now context-dependent, not automatic.

The edge now comes from:

1. Seasonality + extreme positioning
2. Seasonality + storage deviations
3. Seasonality + weather anomaly
4. Technical breakdown + seasonal tailwind

Seasonality alone = weak edge
Seasonality + imbalance = real edge
What Still Works in 2025
A. Extreme Storage Divergence Plays
When inventory is far above/below 5-year average, spreads can trend strongly.
B. Volatility Compression → Expansion
NG spreads often coil before major weather shifts.
C. Shoulder Month Mean Reversion
Mar–Apr and Oct–Nov still show structural tendencies.
Risk Profile vs Outright NG
Calendar spreads:

1. Lower margin
2. Lower volatility
3. More stable behavior
4. Cleaner technical structure

But:

1. Moves are slower
2. Profit per contract smaller
3. Requires patience

Since you’re a day trader, note:
NG spreads are generally better for swing traders than intraday scalpers.
Intraday spread liquidity can be thin outside peak hours.
Where Traders Go Wrong

1. Trading seasonality without checking storage
2. Ignoring COT positioning
3. Trading spreads during low liquidity
4. Assuming historical average = future outcome

Bottom Line
Is NG seasonal spread trading still profitable?
Yes — but only when combined with fundamental imbalance or positioning extremes.
It is no longer a “set and forget” edge.
If you’d like, I can:

1. Break down the most statistically reliable NG seasonal spread
2. Or compare NG spreads vs DAX / NQ intraday trading (since you trade those)

I am still holding one Apex PA account and holding trading until Apex resumes and allows gold trading again. 180 days of inactivity is allowed for PA accounts. The other accounts are all gone trading NQ. Actually I intentionally over leveraged them because I wanted to pass within a month. From the second month, only 50% discount is applied which I don't think worth paying. I purchase prop firm accounts only when they provide a decent discount. I scalp gold based on JMA and basic structure. It works very nicely until it stops working. The most important skill I think is to know when to step away. Choppy, consolidating days should be avoided unless you are a mean reversion based trader. It seems Steve's strategy is also not very vulnerable to choppy price actions because his strategy is not based on breakout momentum. Actually, breakout based traders have a huge disadvantage.

Gold was consistently choppy during this whole week. I have avoided most days but made some losses. Hope next week will be better.


In terms of gold futures market (GC), how often and long market condition changes between trendy and choppy?


In the gold futures market (GC) traded on COMEX (part of CME Group), the shift between trending and choppy (range-bound) conditions happens regularly — but not on a fixed schedule.
Since you’re a day trader, I’ll break this down both structurally and practically.
Big Picture: Macro Regime Changes (Weeks to Months)
Gold tends to trend strongly when driven by:

1. Interest rate cycle shifts (Fed policy)
2. USD strength/weakness
3. Inflation expectations
4. Geopolitical stress

Typical behavior:

1. Strong trending phases: 1–6 months
2. Range-bound/choppy phases: 2–8 weeks

Example:

1. 2020 COVID rally → multi-month trend
2. 2021 large portions → choppy consolidation
3. 2024 breakout → sustained momentum phase

Gold often trends during macro uncertainty, and chops during policy clarity or rate stabilization.
Swing-Level Rotation (Daily Chart)
On the daily timeframe:

1. Clean directional trends: typically 2–6 weeks
2. Choppy consolidation: 1–4 weeks

Gold frequently:

1. Breaks out
2. Runs impulsively
3. Then compresses in volatility
4. Then breaks again

It breathes in expansion–contraction cycles.
Intraday (Most Relevant for You)
For GC day trading, regime shifts happen much faster.
On 5–15 minute charts:

1. Clean trend days: 2–4 days in a row
2. Choppy rotation days: often 1–3 days
3. Sometimes regime flips within the same day

Weekly structure tendency:
Gold often:

1. Trends strongly during 1–2 sessions
2. Rotates/chops 2–3 sessions
3. Then expands again

But during high-volatility macro weeks (CPI, FOMC), it may trend multiple days consecutively.
Volatility Cycles (Important)
Gold alternates between:

1. Volatility expansion → Trend
2. Volatility compression → Chop

This cycle typically resets every:

1. 3–10 trading days (intraday traders)
2. 2–4 weeks (swing traders)

Statistically (Approximate Observation)
Across years of GC behavior:

1. ~40–50% of days are rotational/choppy
2. ~30–35% are directional but moderate
3. ~15–25% are strong trend days

True “clean trend days” are less common than traders think.
What Triggers Regime Shifts?
Gold flips from chop → trend when:

1. Dollar index breaks structure
2. Real yields shift
3. Major macro release surprises
4. Liquidity vacuum above/below range extremes

And trend → chop when:

1. Key level is reached
2. News is priced in
3. Volatility contracts

Practical Answer for a Day Trader
In GC, expect:

1. 2–5 trading days before character shifts
2. Intraday structure shifts every 1–3 sessions
3. Larger regime change every 3–6 weeks
4. Decreasing average true range

In this week, gold breakout trading was relatively easy. At least easier than last week. Look at the following 15M chart. Swings are natural without much chops.



Price action was cleaner without many traps or fakes. Even a simple, classical MA crossover strategy would have worked during this week. This is the time when people mistakenly think they figured out how to trade and purchase many prop firm accounts. Alas, good days don't last long. Soon we will have gloomy days and people will blow out their accounts. In my opinion, the key to success for at least breakout based day traders is their skills to filter out bad days and step away until good days come back. Volatility comes and goes like breathing. I am seeing if I can get some early clues from multiple ATRs....


It looks like good trends usually start when ATR is bottom. I will use multiple periods of ATR to find bottom and use MAs to filter entries. Is this good idea?

That’s actually a very solid volatility-expansion concept
What you're describing is essentially:
Volatility contraction (low ATR) → Volatility expansion (trend move)
filtered by moving averages for direction
For a day trader like you, this makes structural sense. Let’s break it down properly.
Why ATR Bottom → Trend Makes Sense
ATR measures range expansion.
When ATR is at multi-period lows:

1. Market is compressed
2. Order flow is balanced
3. Breakout energy is building
4. Stops cluster outside range

When expansion begins:

1. Volatility expands
2. Range widens
3. Trend legs often begin

This is the classic:
Compression → Expansion cycle
It works especially well in:

1. GC (Gold futures)
2. NQ
3. Crude Oil
4. London/NY session opens

Using Multiple ATR Periods (Very Smart)
Instead of one ATR, using:

1. Short ATR (5–14)
2. Medium ATR (20–50)
3. Long ATR (100–200)

is better because:
ATR StateMeaning
Short low, long highPullback inside trend
All lowTrue compression
Short rising firstEarly expansion
All risingStrong trend underway
The best breakout often occurs when:
Short + Medium + Long ATR are near multi-session lows
That’s when the market is truly compressed.
Adding MA Filter (Very Important)
ATR alone does NOT tell direction.
Using MAs:

1. 20 / 50 EMA for short-term bias
2. 100 / 200 EMA for structural bias
3. Or your JMA rainbow

You could define:
Long setup example

1. ATR(14), ATR(50), ATR(100) near lowest 20% of last X bars
2. Price above 50 MA
3. MA slope up
4. Break of range high

That avoids fake breakouts in chop.
The Hidden Danger
Low ATR does NOT always lead to trend.
Sometimes:

1. It leads to low volatility drift
2. Or fake breakout then revert
3. Or remains compressed for long time

So you need:
Trigger
Don’t enter just because ATR is low.
Wait for:

1. Range breakout
2. Volume expansion
3. Structure break
4. Strong impulse candle

Even Better Version (Pro-Level Adjustment)
Instead of absolute ATR bottom, use:
ATR Percentile
Example:
ATR < 15th percentile of last 100 bars
That’s more adaptive.
Or:
ATR Squeeze Ratio
ATR(14) / ATR(100)
When ratio is extremely low → compression state.
That avoids fixed thresholds.
Where This Works Best
For GC (since you trade futures):
Best times:

1. Pre-London compression → London breakout
2. Asia range → NY expansion
3. Post-news volatility compression

Worst times:

1. Midday US session
2. After big news spike (ATR already expanded)

Will It Work?
Yes — if:
You wait for breakout confirmation
You avoid overtrading compression
You align with session timing
You use structure, not just indicators
It’s a volatility regime strategy, not a signal strategy.
My Honest Opinion
For a structured trader like you (using JMA rainbow, EOM, etc.),
this is actually a better foundation than pure MA crossover systems.
You’re moving toward:
Regime detection → Direction filter → Trigger
That’s professional thinking.
If you want, I can help you:

1. Design exact rule set
2. Turn it into Quantower logic
3. Or stress-test the concept structurally for GC

Do you want this for scalping (1–5 min) or intraday swings (5–15 min)?

Why fully autonomous AI day trading is still limited?


Short answer: markets are messy, adaptive systems, and fully autonomous AI still struggles with that kind of environment—especially on short timeframes like day trading.
Here’s what actually limits it:
1) Non-stationary markets (the rules keep changing)
Most AI (especially machine learning) assumes patterns repeat.
In reality:

1. Market behavior shifts with macro cycles, liquidity, policy, and participant behavior
2. A strategy that works this month can break next month

This is a classic **non-stationarity problem.
AI trained on past data often overfits and fails when conditions change.
2) Reflexivity: the market reacts to the model
Markets aren’t passive.

1. If a strategy works, others copy it
2. That edge disappears or even reverses

This is tied to reflexivity (popularized by George Soros)
Your AI’s success can literally destroy its own edge.
3) Data limitations (especially for intraday)
AI needs:

1. Huge amounts of clean, labeled data
2. Consistent structure

But day trading data has issues:

1. Noisy (random price movement dominates)
2. Sparse true “signals”
3. Expensive high-quality order flow data

The signal-to-noise ratio is terrible.
4) Black swan & regime shocks
AI struggles with events it hasn’t seen before.
Examples:

1. COVID-19 market crash
2. Flash Crash

These events:

1. Break historical correlations
2. Cause extreme volatility spikes

Humans can adapt conceptually; AI often fails catastrophically.
5) Execution is harder than prediction
Even if AI predicts direction correctly:

1. Slippage
2. Latency
3. Liquidity constraints

matter a lot.
Institutional firms invest heavily in:

1. Co-location servers
2. Ultra-low latency systems

Retail-level AI can’t compete here.
6) Risk management is still a weak point
This is one of the biggest issues.
AI models:

1. Optimize for returns in training
2. Often ignore tail risk or drawdowns

Without strict constraints, they can:

1. Over-leverage
2. Blow up quickly

Many AI systems look great in backtests and fail live.
7) Reinforcement learning isn’t mature enough (yet)
RL is often seen as the “holy grail” for autonomous trading.
But in markets:

1. Rewards are delayed and noisy
2. Exploration is expensive (real money)
3. Environments are adversarial

Training stable, reliable agents is extremely difficult.
8) Regulation and control concerns
Firms don’t fully trust autonomous systems:

1. Compliance requirements
2. Need for human oversight
3. Risk of runaway behavior

Most systems are semi-automated, not fully autonomous.
The real reason (boiled down)
Fully autonomous AI works best in:

1. Stable environments
2. Clear rules
3. Repeatable patterns

Day trading is the opposite:
dynamic, adversarial, noisy, and constantly evolving
What actually works today
Instead of full automation, the industry standard is:
Human + AI hybrid

1. AI: scanning, execution, optimization
2. Human: context, risk judgment, regime awareness

Even top quant firms rarely rely on a single fully autonomous system.
Straight talk (important for you as a trader)
If fully autonomous AI day trading actually worked reliably:

1. Markets would become extremely efficient
2. Edges would vanish quickly

The fact that:

1. discretionary traders
2. simple strategies (VWAP, EMA, order flow)

still work at times
tells you AI hasn’t “solved” trading.
If you want, I can show you where AI actually gives a real edge in day trading (and where it’s mostly hype)—that’s the part most traders misunderstand.

Hi Justin,
I've never really looked at AI for Trading but I guess it would be good at spotting Specific Price Patterns?

Hi, Steve
I don't really know much about AI but anyway it's good to know there's still some small room for mere human traders like you and me.... If AI will become really really matured in the future, I guess the conventional markets we see today would disappear one by one..... If every resources can be distributed highly efficiently and fairly by AI, why do we need 'markets' in the end?

If AI is fully matured, will conventional market systems disapear?


Short answer: no—markets probably won’t disappear. But they could look very different.
Even with highly advanced AI, the core reason markets exist doesn’t go away: scarcity and differing preferences. As long as resources (time, land, energy, attention, capital) are limited and people (or organizations) value things differently, some form of exchange system is needed. That’s basically the foundation of economics.
What will change is how markets operate:
1. Price discovery may become near-instant
AI could process information faster than any human or current algorithmic systems. In theory, this pushes markets closer to the idea of efficient market hypothesis—where prices instantly reflect all available information. That would reduce many traditional trading edges.
2. Human traders lose most advantages
In day trading (which you’re familiar with), discretionary decision-making would struggle against AI systems that:

1. ingest global data in real time
2. model probabilities better
3. execute with zero latency

At that point, human trading becomes more niche (intuition, behavioral anomalies, or very long-term plays).
3. Markets might shift from “trading” to “allocation”
Instead of people speculating, AI systems could continuously allocate capital across the economy in an optimized way—kind of like a dynamic, real-time planning system rather than a chaotic marketplace.
4. Volatility could either shrink—or spike
Two competing possibilities:

1. If AI systems converge on similar models → smoother, more stable markets
2. If they compete with different strategies → rapid feedback loops and flash crashes on steroids

We already see hints of this with high-frequency trading.
5. Ownership and incentives still matter
Even with AI, you still need answers to:

1. Who owns resources?
2. Who gets profits?
3. Who decides risk?

Those questions keep markets (or something very similar) alive. Without them, you’re moving toward a centrally planned system—which has its own well-known limitations.
The real shift isn’t “markets disappear” — it’s:

1. fewer inefficiencies
2. fewer human participants at the execution level
3. more AI-vs-AI competition
4. humans moving to meta-level decisions (designing systems, setting goals)

If AI ever becomes so advanced that scarcity itself is largely solved (e.g., energy, production, logistics become near-free), then markets could shrink dramatically. But that’s a much bigger leap than just “mature AI.”
If you want, I can break this down specifically from a day trading perspective—like what edges might still exist in an AI-dominated market.

Trade following the VWAP and EMA direction. Is this a good idea?


Short answer: it’s a solid baseline idea, but by itself it’s not enough to be consistently profitable.
Using VWAP + EMA alignment works because you’re essentially trading trend + fair value bias:

1. VWAP → tells you where institutional “average price” sits (intraday bias)
2. EMA(s) → show short-term momentum/trend

When both point the same way, you’re trading with order flow instead of against it. That’s good. But the edge comes from how you use it, not just the concept.
Where it actually works well
You’ll see this combo perform best in:

1. Strong intraday trends
2. High volume sessions (open, news, momentum stocks)
3. Clean structure (higher highs / higher lows or vice versa)

Example:

1. Price above VWAP
2. VWAP sloping up
3. EMA stack (e.g., 9 > 20 > 50)
   → Pullback to EMA or VWAP → continuation

That’s a legitimate setup.
Where it fails (important)
This is where most traders lose money with it:

1. Choppy / sideways markets → constant fake signals
2. Price crossing VWAP repeatedly (no clear bias)
3. EMAs flattening → no momentum edge

If you blindly follow “VWAP + EMA direction,” you’ll get chopped up.
What actually makes it work
You need at least one more layer:
1. Context (most important)
Ask:

1. Is this a trend day or range day?
2. Is price accepting above/below VWAP, or just wicking through?

No context = no edge.
2. Entry trigger (don’t enter just because aligned)
Better triggers:

1. Pullback + rejection (wicks, strong candle close)
2. Break + retest of VWAP/EMA
3. Volume confirmation on continuation

3. Risk management
This strategy needs tight control:

1. Stop below VWAP or structure
2. Avoid chasing extended moves far from VWAP

4. Confluence (this is where pros differ)
Add one of:

1. Key levels (previous high/low, premarket levels)
2. Volume spikes
3. VWAP bands / standard deviations
4. Time of day (open vs midday matters a lot)

Reality check (important)
A lot of beginners think:
“If VWAP and EMA agree → buy/sell”
That’s too simplistic. Institutions don’t trade that way.
Think of VWAP + EMA as:
a filter, not a signal
Bottom line

1. Good directional framework
2. Not a standalone strategy
3. Dangerous in chop without context

If you want, I can break this into a precise rule-based system (entries, exits, invalidation) tailored for your day trading style.

Restarted Lucid Eval as a practice before I go to live. Slow progress with small account. I find trade following VWAP is a bit easier. 3 trades today. One loser.





Once I feel confident after a couple of months of test, I will trade my own fund. I am not greedy. Average $500 per day is really all I need.

Took one trade moments before. Only one so far. Following the VWAP....

One quick trade in this Asian session. I am going to bed and coming back on last Europe session. I work at night in my shop and have sleep during afternoon.

BE so far... I should've taken out quick and run.....

Second one is loss trading. NY session is more difficult indeed....

Price is stuck between two MAs on 4H chart. It may be difficult to trade under this condition.....