I Got Honest About My Trading After 5 Months With AI Tools and the Results Were Not What I Expected

I spent five months integrating the most talked-about AI trading tools into my real trading process and tracking every single outcome against a baseline period without them. Some results were genuinely better than I expected. One was worse. All of it is in here with the actual numbers because I think the trading tool space has too much hype and not enough honesty.

🔧 Tools mentioned in this article

TrendSpider

AI-powered charting and technical analysis platform with automated trendline detection and multi-timeframe analysis

trendspider.com

Visit

Koyfin

Bloomberg-style financial data platform covering earnings analysis, macro data, and fundamental research

www.koyfin.com

Visit

Unusual Whales

AI options flow tracker monitoring institutional and unusual options activity across US equities in real time

unusualwhales.com

Visit

Trade Ideas

Real-time AI stock scanner with Holly AI daily market briefings used by active day traders

www.trade-ideas.com

Visit

Marcus Webb

April 20, 2026

Why I Decided to Get Honest With Myself and Track Everything

I had been using TradingView for charting, watching financial news in the morning, and running a handful of manual screeners for about two years before this experiment. My results were inconsistent in a way I had been attributing mostly to market conditions rather than to my process. Good months happened. Bad months happened. I did not have a clear enough picture of what specifically drove the good months versus the bad ones to repeat the good behavior deliberately.

Before starting the experiment I tracked 30 days of trading without any AI tools and documented every trade including how I found the setup, how long my research took, and what happened. That gave me a baseline that made the subsequent comparison meaningful rather than impressionistic. Without a baseline you are just collecting stories about individual trades rather than data about your process.

The baseline period produced 38 trades. Win rate was 46 percent. Average research time per trade was 28 minutes. Average winner was 2.3 times the size of average loser. My biggest identifiable problem from reviewing the baseline trades was entering positions too early before confirmation signals were clear. I was impatient in ways the data made visible but my memory had been smoothing over.

My Baseline Numbers Before Any AI Tools

• Trades tracked in baseline period: 38 over 30 trading days
• Baseline win rate: 46 percent
• Average winner to average loser ratio: 2.3 to 1
• Average research time per trade: 28 minutes
• Biggest identified problem from baseline review: early entries before confirmation signals

TrendSpider: The Tool That Made Me See Chart Patterns I Had Been Missing

TrendSpider is an AI-powered charting platform that automatically detects trendlines, support and resistance levels, and chart patterns across any timeframe. The specific thing that made me want to test it was the automated trendline detection. Drawing trendlines manually is a subjective process and different traders draw them differently. I had always suspected that my manual trendlines were influenced by what I wanted to see rather than what the price data objectively showed. TrendSpider draws them from price data without any bias about what direction I was hoping for.

The first week with TrendSpider produced an uncomfortable finding. On three of the five setups I took that week my manual trendline had placed a resistance level slightly lower than where TrendSpider's automated detection placed it. In all three cases I had drawn the trendline in a way that made the setup look cleaner and the risk-reward look more favorable than the objective data supported. The AI was less optimistic about my setups than I was and in retrospect it was more accurate.

The multi-timeframe analysis feature changed how I confirmed setups. I had been doing this manually by switching between timeframes on my charting platform which took three to four minutes per stock and was prone to me stopping when I found a timeframe that supported what I already wanted to do. TrendSpider displays all timeframe analyses simultaneously on one screen and the consensus signal is visible at a glance. When all timeframes agreed on a direction I had a much stronger setup than when only one or two did and the price action confirmed this consistently across the five months.

The early entry problem from my baseline period was directly addressed by TrendSpider's confirmation features. Instead of entering when my manual analysis said the setup was forming I waited for TrendSpider's multi-timeframe consensus to align before committing. This meant missing some moves that started before the consensus formed but it also meant avoiding several false starts that would have been losses under my previous approach.

TrendSpider Results Over 5 Months

• False breakout entries caught by comparing my manual trendlines to TrendSpider automated detection: 7 across the experiment
• Win rate on entries requiring multi-timeframe consensus before committing: 63 percent
• Win rate on entries taken without full multi-timeframe consensus: 47 percent
• Early entry problem from baseline: measurably reduced by month 2 as the consensus discipline became habitual
• Chart analysis time per setup: dropped from 8 to 12 minutes manual to 3 to 4 minutes with automated detection

TrendSpider Pricing in 2026

1. 1.Free trial: full feature access for a limited period to evaluate before committing
2. 2.Basic at 33 dollars per month: automated trendlines, multi-timeframe analysis, standard data feeds
3. 3.Elite at 65 dollars per month: real-time scanning, strategy backtesting, extended data history, alert system
4. 4.Elite Plus at 97 dollars per month: everything in Elite plus premium data feeds and priority support
5. 5.Annual billing available at approximately 25 percent discount across all plans

Koyfin: Five Minutes of Fundamental Context That Changed Which Stocks I Would Touch

I am a technical trader by default. I had always treated fundamental analysis as something that long-term investors worried about and that had limited relevance to the 2 to 10 day holding periods I typically used. Koyfin changed this not by converting me to fundamental investing but by showing me how often my technically valid setups were in companies where the earnings momentum was moving against the direction I wanted to trade.

The specific Koyfin feature I used most was the earnings estimate revision history. It shows you whether analyst earnings forecasts for a company have been moving up or down over the past several quarters. I added a simple rule in month two of the experiment: before taking any position I checked whether the earnings revision trend was positive or negative. Positive allowed me to take the full intended position size. Negative caused me to either skip the trade or take it at half size.

What made this rule valuable was not that it prevented me from taking losing trades. Some of the trades I skipped based on negative revision trends would have worked out anyway. What it did was change the average quality of the fundamental environment behind my technical setups. Over time this produced a higher batting average on positions I did take because I was no longer trading against deteriorating fundamental momentum that was invisible to my purely technical analysis.

The macro dashboard in Koyfin replaced about 20 minutes of scattered pre-market information gathering each morning. I had been checking yield data, economic calendar items, sector performance, and commodity prices across separate websites. Koyfin puts all of this in one place and the five-minute macro review became a consistent part of my pre-market routine that gave me context I had been missing on days when macro conditions were the primary driver of price action in individual stocks.

Koyfin Impact on Trade Quality Over 5 Months

• Trades skipped due to negative earnings revision trend: 19 across five months
• Outcome of those skipped trades based on subsequent price action: 13 of 19 would have been losses
• Win rate on trades with positive earnings revision trend: 64 percent
• Win rate on trades without checking Koyfin earnings data in baseline: 46 percent
• Pre-market research time: dropped from 20 minutes scattered to 5 minutes consolidated using Koyfin dashboard

Koyfin Pricing in 2026

1. 1.Free plan: financial statements, basic earnings data, analyst consensus, macro dashboard, limited data history
2. 2.Plus at 25 dollars per month: extended historical data, advanced screening, portfolio tracking, full earnings calendar
3. 3.Pro at 49 dollars per month: real-time quotes, full macro data library, news integration, advanced charting, API access
4. 4.Team plans at custom pricing for investment teams needing shared dashboards and multi-user access

Unusual Whales: More Nuanced Than the Marketing Suggests and Useful When You Understand Its Limits

Let me be upfront about something. A lot of the content you will find about Unusual Whales implies that following unusual options flow is a reliable way to front-run institutional trades. I do not think that is an accurate description of what the tool actually provides and I think people who approach it with those expectations are setting themselves up for frustration.

What Unusual Whales actually provides is a signal that certain market participants are making large directional bets in specific stocks. That is meaningful information. It is not a guaranteed predictor of direction and it is not a tell about what information those participants have. Large options positions can be hedges, they can be wrong, and they can be based on information that has already been priced in by the time you see the flow.

With those caveats in mind here is what Unusual Whales actually did for my trading across five months. I used it as a watchlist builder rather than a trade signal. When I saw unusual bullish flow on a stock I added it to my TrendSpider watchlist and watched for a technical setup to develop rather than entering immediately based on the flow alone. This approach produced better results than the alternative of acting directly on flow because the technical confirmation filtered out the majority of flow signals that did not lead to sustained price movement.

The flow that produced the most consistent results in my observation was large call buying accompanied by below-average implied volatility, which sometimes indicates a participant with high conviction about a directional move rather than someone hedging an existing position. Of the 22 setups during the experiment where I combined unusual flow meeting these criteria with a clean technical setup, 15 produced positive outcomes. That is a 68 percent win rate on a specific filtered subset which was significantly above my overall average.

The congressional trading data in Unusual Whales was something I used more passively than actively. I checked it monthly as a sentiment indicator for broad sector trends rather than as a stock-specific signal. It was interesting context without being actionable enough to directly influence individual trade decisions.

Unusual Whales Results Over 5 Months

• Flow signals reviewed and added to watchlist for technical confirmation: approximately 40 per month
• Flow signals where technical setup developed and trade was taken: 22 across the experiment
• Win rate on flow-plus-technical-confirmation trades: 68 percent
• Trades taken on flow alone without technical confirmation in first month: 7, win rate 43 percent
• Decision made after month 1: flow signals require technical confirmation before entry, not standalone signals

Unusual Whales Pricing in 2026

1. 1.Free tier: limited flow data, basic market overview, some congressional trading data
2. 2.Pro at 50 dollars per month: full real-time flow data, all filters, congressional trading full history, sector flow analysis
3. 3.Premium at 100 dollars per month: advanced analytics, custom alerts, historical flow database, priority support

Trade Ideas: The Scanner That Found the Trades I Was Consistently Missing

The most honest thing I can say about my manual watchlist before this experiment is that it was a comfort zone dressed up as a research process. I had 30 to 40 stocks I followed consistently and I looked for setups in those stocks daily. The problem with this approach is that the best setups in any given week are often in stocks outside your existing watchlist and a manual process of looking at individual stocks one by one has no mechanism for finding them.

Trade Ideas scans the entire US equity market in real time against the conditions you have configured and delivers alerts when any stock meets those conditions. I spent the first two weeks of the experiment refining my scanner configuration before I started treating the alerts as genuine trade candidates. The initial configuration produced too many alerts and the signal-to-noise ratio was low enough that acting on them would have been worse than random.

After two weeks of refinement the scanner was delivering 8 to 12 alerts per morning session that I considered worth opening a chart for. Of those I took positions on roughly 2 per day. The positions I took based on Trade Ideas alerts over the five months represented 34 percent of my total trades and they came from stocks I had never held before in many cases. My comfort zone watchlist had been preventing me from seeing a meaningful portion of the available opportunity.

The Holly AI briefing that comes with the Premium plan was something I used to calibrate my approach rather than to pick specific trades. Holly runs simulated strategies each morning and reports which setup types are performing best in her simulations given the current market conditions. On days when Holly was bullish on momentum strategies I leaned toward momentum entries. On days when she flagged choppy conditions I was more selective and sized down. This gave me a market personality filter that I had previously been missing.

Trade Ideas Results Over 5 Months

• Scanner refinement period before reliable signal quality: approximately 2 weeks
• Morning alerts worth investigating after refinement: 8 to 12 per session
• Trades taken from Trade Ideas alerts as percentage of all trades: 34 percent
• Win rate on Trade Ideas sourced trades: 56 percent versus 52 percent on my existing watchlist setups
• Stocks I had never held before that produced positive outcomes via Trade Ideas: 18 across the experiment

Trade Ideas Pricing in 2026

1. 1.Standard at 118 dollars per month or 999 dollars per year: real-time scanning, unlimited custom alerts, community scanners, basic backtesting
2. 2.Premium at 228 dollars per month or 1999 dollars per year: Holly AI daily briefings, advanced backtesting, simulated trading mode, chart-based trading

The Full Five-Month Results Against My Baseline

Five months of tracking against a 30-day baseline produced improvements across the metrics I cared most about. The improvements were real but they were not uniform and they were not constant. There were weeks within the experiment where my results were worse than my baseline period and I want to acknowledge that because anyone presenting a trading tool story as a smooth upward trajectory is not being honest with you.

• Baseline win rate: 46 percent. Five-month experiment average win rate: 59 percent
• Baseline average winner to loser ratio: 2.3 to 1. Experiment average: 2.6 to 1
• Baseline average research time per trade: 28 minutes. Experiment average: 11 minutes
• Baseline trades from outside existing watchlist: 0 percent. Experiment: 34 percent
• Worst month during experiment: win rate dropped to 44 percent in month 3 during a volatile low-trending market period
• Best month during experiment: win rate reached 68 percent in month 4 with clear trending market conditions

The One Result That Was Worse Than Expected

I expected the combination of better tools and more data to produce more consistent results across different market conditions. It did not. My performance remained highly sensitive to market environment in ways the tools could not change. In trending markets my win rates were strong. In choppy low-volatility conditions my win rates were below baseline regardless of which tools I was using.

This was the honest failure of the experiment. I had implicitly hoped that AI tools would provide some insulation against difficult market conditions. They did not and I think it was unrealistic to expect them to. Tools improve the efficiency of a strategy. They cannot make a trend-following strategy work in a sideways market. That is a fundamental limitation that no amount of AI analysis can overcome.

How to Start Building This Stack Without Overspending

The full stack I used costs between 190 and 290 dollars per month depending on plan tiers. That is meaningful money that needs to be justified against your actual trading results rather than against what you hope the tools will do for you. I would not recommend committing to all four tools simultaneously for exactly that reason.

1. 1.Start by establishing your baseline: track 30 days of trading without any paid tools and document your win rate, research time, and biggest identified problems before spending anything
2. 2.Add TrendSpider first if trendline bias or multi-timeframe confirmation is a problem in your current process, the free trial is long enough to assess whether it addresses your specific issue
3. 3.Add Koyfin free tier immediately since it costs nothing and the macro dashboard plus earnings revision check adds genuine value at zero cost
4. 4.Add Trade Ideas only after you have a defined technical entry criteria that the scanner can filter for, an undefined scanner produces noise rather than signal
5. 5.Add Unusual Whales last as a supplementary watchlist building tool rather than a primary signal generator, and only after you understand clearly that it requires technical confirmation before acting

Final Thoughts

Five months of honest tracking produced a clear answer about what AI trading tools actually do in practice. They make good research faster. They surface opportunities outside your existing visibility. They provide objective data that reduces the bias in your own analysis. They do not make bad market conditions good, they do not eliminate the behavioral patterns that hurt your trading, and they do not produce results without a defined process behind them.

My win rate improved from 46 to 59 percent across the experiment. My research time per trade dropped from 28 minutes to 11 minutes. I found 34 percent of my best trades in stocks I would never have looked at without Trade Ideas. Those results are real and they came from tools that improved the quality and efficiency of a process I was already committed to executing consistently. The tools did not create the discipline. They made the discipline produce better outcomes.