As of 12/20/2024
Indus: 42,840 +498.02 +1.2%
Trans: 15,892 +32.54 +0.2%
Utils: 986 +14.76 +1.5%
Nasdaq: 19,573 +199.83 +1.0%
S&P 500: 5,931 +63.77 +1.1%
|
YTD
+13.7%
0.0%
+11.9%
+30.4%
+24.3%
|
44,200 or 41,750 by 01/01/2025
16,100 or 17,700 by 01/01/2025
1,050 or 975 by 01/01/2025
20,500 or 19,300 by 01/01/2025
6,100 or 5,775 by 01/01/2025
|
As of 12/20/2024
Indus: 42,840 +498.02 +1.2%
Trans: 15,892 +32.54 +0.2%
Utils: 986 +14.76 +1.5%
Nasdaq: 19,573 +199.83 +1.0%
S&P 500: 5,931 +63.77 +1.1%
|
YTD
+13.7%
0.0%
+11.9%
+30.4%
+24.3%
| |
44,200 or 41,750 by 01/01/2025
16,100 or 17,700 by 01/01/2025
1,050 or 975 by 01/01/2025
20,500 or 19,300 by 01/01/2025
6,100 or 5,775 by 01/01/2025
| ||
Updated with new performance information on 11/11/24.
One-Day Reversal, Bottom
|
Characteristic | Discussion |
3 bars | The pattern is composed of one bar (the middle of the three), but for identification, I use three bars, one day before to one day after the one-day reversal. |
Bottom | Look for the pattern in a short-term down trend. In other words, wait for an upward breakout (a close above the top of the middle bar). |
Open and close | The open and close on the one-day reversal must be within 25% of the intraday high. |
Surrounding days | The low price of the two adjacent bars must be above the mid point of the one-day reversal. This should make the one-day reversal bar stand alone, like a tree atop a peak (only inverted). |
Tall | The one-day reversal should be at least as tall as the one-month average height of other price bars. |
Volume | High volume should be present on the one-day reversal. However, I excluded this requirement since the pattern is rare enough without it. |
Trading Tactic | Explanation |
Reversal | The pattern is supposed to act as a reversal of the downtrend, so only trade those that reverse the downtrend. |
Buy | After identifying the 3-bar pattern, place a buy-stop a penny above the top of the middle price bar. |
Measure rule | The one-day reversal fulfills the measure rule 73% of the time (bull market). That is, measure the height of the pattern and add it to the high price to get an upward target. |
For the following statistics, I used 1,160 stocks, starting from January 1990 to March 2013, but few stocks covered the entire range. All stocks had a minimum price of $5. There were two bear markets in the 2000s (as determined by the S&P 500 index), from 3/24/2000 to 10/10/2002 and 10/12/2007 to 3/6/2009. Everything outside of those dates represents a bull market.
For each one-day reversal, I found when the trend started and when it ended. To find the trend peak or valley, I found the lowest valley and highest peak within plus or minus 10 days (21 days total) each, before the one-day reversal and the same peak/valley test after the one-day reversal. The closest valley or peak before the one-day reversal is where the trend began. The closest peak or valley after the one-day reversal is where the trend ended. I compared the peak or valley to the average of the highest high and lowest low price of the one-day reversal pattern.
The 10-bar peak or valley number tends to find major turning points on the daily charts.
I measured performance from the day after the pattern ended to the nearest trend peak or trend valley.
To determine the inbound price trend (I was looking for a down trend), I used linear regression on the average of the high-low prices in the five days before the pattern. That caught the short-term trend.
Market | 5% Failure | Average Rise |
Bull | 39% | 8% |
Bear | 31% | 10% |
Table 1 lists failure rates, sorted by market condition along with the average rise. Since the one-day reversal acts as a reversal of the downward trend, I looked for an upward breakout.
A failure occurs when the stock fails to rise more than 5%.
The failure rates may appear high, but that's typical for short-term patterns like the one-day reversal. The highest failures occur in a bull market: 39% fail to see price rise at least 5%. The average rise is just 8%.
Market | Success |
Bull | 73% |
Bear | 65% |
Table 2 shows how often the measure rule works. Use the measure rule to estimate of how far price is likely to drop.
To do this, measure from the highest high to the lowest low in the pattern to get the height. Add the height to the highest high to get the target.
The best performance of the measure rule occurs in a bull market, with 73% of patterns reaching their target. This makes sense since a bear market is trending lower, making it harder for an upward breakout to outperform in a stock (it fights the downtrend).
Market | Bull | Bear |
Net profit/loss | $61.35 | $(92.17) |
Wins | 56% | 45% |
Winning trades | 2,193 | 403 |
Average gain of winners | $705.07 | $711.62 |
Losses | 44% | 55% |
Losing trades | 1,752 | 496 |
Average loss | ($744.40) | ($745.25) |
Average hold time (calendar days) | 27 | 14 |
Table 3 shows the performance based on 5,532 trades using $10 commissions per trade ($20 round trip), starting with $10,000 per trade. No other adjustments were made for interest, fees, slippage and so on.
The results are sorted by bull or bear market. The trades used the same setup as listed in One-Day Reversal, Bottom, Performance Statistics.
Here's the setup.
For example, in a bull market, the net gain was $61.35 for all trades. The method won 56% of the time and there were 2,193 winning trades. The average gain of winning trades was $705.07.
Forty-four percent, or 1,752 trades were losers. They lost an average of $744.40.
The average hold time was 27 calendar days.
Notice how the gains and losses were pegged near 7% (of the $10,000 investment), which is how the test was setup.
The figure shows a one-day reversal pattern in Alaska Air (ALK) on the daily scale, at A.
Price drops into the one day reversal. The open and closing prices are near the top of the pattern. The day is an unusually tall one, too.
The next day, the stock closes above the top of the one-day reversal. Buy at the open the following day, B.
When the stock climbs to the sell target, C, close out the position. If the stock tumbled instead of rising, a stop placed 7% below the buy price would have closed out the trade for a loss.
The following tests are different from the ones above.
In the discussion that follows, I use twice the height of the ODRB (the middle price bar of the 3-bar pattern) as a price target to sell, and a stop loss order placed a penny below the day to limit adverse moves.
I show an example trade in 3M stock. I highlighted the ODR pattern. Entry occurs as shown with a buy-stop placed a penny above the top of the middle price bar in the ODR. Because we wait one day after the middle price bar to be sure the ODR identifies as a valid pattern (meaning the bottom spike is well below the two adjacent bars), entry is delayed by (at least) one day.
The target is twice the height of the ODR day added to the top of the ODR.
A stop loss order is placed a penny below the bottom of the ODR. In this example, price fails to rise far enough to trigger a profitable exit. Instead, the stop loss order triggers for a losing trade.
Testing for the ODR is different than for other small patterns. Why? The one-day reversal is a one day pattern but I also use the two adjacent days to make sure the ODR sticks out. However, the buy and sell points still trigger on the middle day (the ODR itself) even though entry is delayed by a day. The benchmark follows this scheme by delaying entry until the day after the 3-bar pattern completes (then price is compared to the ODR day and a breakout buy consummated if necessary).
I used a target exit placed twice as high as the height of the ODRB day (the middle price bar of the three). I placed a stop loss a penny below the bottom of the ODR day. For additional methodology details, see the link.
Tables 4, 5, and 6 show results for bull markets with an upward breakout and a downward inbound price trend. I used 497 stocks in the test. I show the 1-day delay entry scheme for the benchmark, but also the non-delayed entry for the benchmark. For the non-delay column, I placed a buy stop a penny above the top of the pattern for entry, and a stop loss order a penny below the price bar for a stop exit. The target exit remains as twice the height of the price bar.
Metric | ODRB in Downtrend | 1-Bar Delay Benchmark | No Delay Benchmark |
Trades | 2,097 | 5,302 | 5,024 |
Average profit/loss per trade | $84.30 | $51.85 | $62.36 |
Win/loss ratio | 52% | 53% | 46% |
Average hold time (days) | 16 | 8 | 8 |
Winning trades | 1,100 | 2,805 | 2,299 |
Average gain of winners | 6% | 4% | 5% |
Average hold time of winners (days) | 19 | 9 | 10 |
Losing trades | 997 | 2,497 | 2,725 |
Average loss | -5% | -4% | -3% |
Average hold time of losers (days) | 15 | 9 | 8 |
The results for the ODRB columns show that it handily outperforms the benchmark (delayed column) in an apples-to-apples comparison, with a profit per trade of $84.30 versus the benchmark's $51.85.
When there is no delay on entry of at least a day, the profit per trade increases from $51.85 to $62.36. This makes intuitive sense. If you're going to have a winning trade, then buy as soon as you can. However, notice that the win/loss ratio drops.
I show a sample trade in U.S. Aerospace & Defense ETF (exchange traded fund, ITA).
The ODRB is the long price spike shown. I placed a stop-loss order a penny below the pattern's low to protect against a big loss, and a buy stop a penny above the high to enter the trade quickly. Because I wait for price bar A to appear well above the ODR day, entry occurs a day later, at the opening price.
The height of the pattern is the highest high minus the lowest low based only on the middle price bar (the ODR day). Multiply that by two and add it to the top of the ODR day. When price reaches the target, sell.
As the chart shows, the ETF reached the target and sold for a profit.
This is the same test as the prior one except I used 94 exchange traded funds (ETFs) instead of common stocks.
Metric | ODRB in Downtrend | 1-Bar Delay Benchmark | No Delay Benchmark |
Trades | 479 | 5,419 | 5,094 |
Average profit/loss per trade | $75.94 | $29.98 | $49.24 |
Win/loss ratio | 61% | 60% | 53% |
Average hold time (days) | 13 | 6 | 6 |
Winning trades | 292 | 3,239 | 2,725 |
Average gain of winners | 4% | 2% | 3% |
Average hold time of winners (days) | 17 | 7 | 7 |
Losing trades | 187 | 2,180 | 2,369 |
Average loss | -3% | -2% | -2% |
Average hold time of losers (days) | 15 | 7 | 6 |
Notice the significantly higher number of trades used in the benchmark. I expect the results from the benchmark columns to be stable (unlikely to change significantly) compared to the ODRB trades.
These trades have some of the highest win-loss ratios I've seen for small patterns. Usually it hovers around 44% but it is 61% for the ODRB. Unfortunately, that doesn't translate into big bucks. Yes, the ODRB clobbers the delayed benchmark by $75.94 to $29.98, but I'd like to see the ODRB higher.
The no-delay benchmark shows a profit almost twice as high as the delayed variety.
This is an example trade in the cryptocurrency AAVE on the daily scale. I highlight the ODRB day.
Entry uses a buy-stop placed a penny above the top of the ODR day. Entry occurs at the opening price that ends the day lower (the price bar turns into a black candle).
A stop-loss order helps minimize the loss with a sell price of a penny below the bottom of the ODR day. The two green lines show the approximate buy stop and stop-loss order locations, but not where they filled.
The currency climbs far enough to reach the target exit.
This is the same test as the prior one except I used 38 crypto currency stocks instead of common stocks.
Metric | ODRB in Downtrend | 1-Bar Delay Benchmark | No Delay Benchmark |
Trades | 25 | 5,796 | 3,443 |
Average profit/loss per trade | $647.53 | $112.00 | $122.37 |
Win/loss ratio | 72% | 54% | 44% |
Average hold time (days) | 7 | 4 | 5 |
Winning trades | 18 | 3,110 | 1,504 |
Average gain of winners | 12% | 8% | 10% |
Average hold time of winners (days) | 8 | 4 | 4 |
Losing trades | 7 | 2,686 | 1,939 |
Average loss | -7% | -6% | -6% |
Average hold time of losers (days) | 9 | 5 | 5 |
The results for the ODRB in crypto land is unreliable due to the low sample count. The results are impressive but can't be trusted. Notice that the no-delay benchmark outperforms the delayed variety (at least in terms of average profit). That's consistent with what we found using other security types.
-- Thomas Bulkowski
Support this site! Clicking any of the books (below) takes you to
Amazon.com If you buy ANYTHING while there, they pay for the referral.
Legal notice for paid links: "As an Amazon Associate I earn from qualifying purchases."
My Stock Market Books
|
My Novels
|
Don't try to have the last word. You might get it