美股投资young
美股投资young
美股科技投资者 ai狂热者 分享每日快讯. all in ai
303Following
742followers
Feed
Livestream
Feed
When volatility doubles, the first thing to do is cut down position size; direction judgment can come later. $SOXX $NVDA
This logic is very straightforward. With the same position size, normally the daily volatility is one or two points, but now it easily swings three to seven points. On 10/30, the semiconductor index dropped 7.2%, on 11/4 it fell 3.13%, and in between there was a day with a 2.63% rise. When volatility doubles and position size remains unchanged, it means the risk has doubled.
Many people make the mistake of using their usual position size to endure the volatility during panic periods. Positions are too heavy, every drop forces a decision, and sleep becomes restless at night. When pressured, people tend to sell at the worst possible time.
Cutting position size and being bearish are two different things. The logic can still be to stay bullish on AI, as the boundaries haven’t been broken, but unit risk must be recalculated based on the current volatility. If volatility doubles, position size should be halved, so the daily account fluctuations remain similar to usual, and judgment stays sharp without being swayed by daily ups and downs.
Add back positions when volatility subsides. Volatility easing itself is a signal that panic is being digested, and adding positions then makes risk more controllable. This is why I always say not to use leverage for long positions. Leverage acts as an amplifier in low volatility but as an accelerator in high volatility, accelerating losses in the wrong direction.
Boundaries remain unchanged: cloud providers cut capital expenditures, DRAM and HBM weaken, and the 10-year US Treasury yield returns above 5%. Direction is determined by boundaries, position size by volatility.
Survive first, then worry about being right.
Since 1999, $SPY has dropped more than 3% in a single day a total of 95 times, and today’s approximately −4.12% drop marks the 95th time; on the second trading day after the previous 94 times, the market rose 57.4% of the time, with an average gain of about +0.39% and a median of about +0.24%. $SPY $QQQ $SOXX $NVDA
These 95 days are almost all clustered in crisis years. In 2008 alone, there were 21 such days, 10 in 2009, 15 in 2020, 8 in 2022, 6 in 2011, and 5 in 2018. Some say the causes behind these were the financial crisis, the Eurozone debt and US debt downgrades, the pandemic shutdown, interest rate hikes and inflation, as well as the 2018 volatility crash and trade war.
Partial list since 2018:
2018/2/5 −4.18%, next day +1.97%
2018/10/10 −3.17%, next day −2.20%
2019/8/5 −3.01%, next day +1.40%
2020/3/11 −4.87%, next day −9.57%
2020/3/12 −9.57%, next day +8.55%
2020/3/16 −10.94%, next day +5.40%
2022/5/5 −3.55%, next day −0.60%
2022/9/13 −4.35%, next day +0.38%
2025/4/3 −4.93%, next day −5.85%
2025/4/10 −4.38%, next day +1.78%
On 2020/3/11, after the drop, the next day fell another 9.57%; on 2025/4/3, after the drop, the next day fell another 5.85%. The probability of a rise the next day is only slightly above half, and the times of continued decline tend to be deeper. Chasing the rebound the next day is not a free lunch.
Today the S&P dropped about 4.12%, and some say the main reason is the panic over AI and data center regulation spreading to the broader market. History can only provide the baseline: volatility tends to cluster, and a big drop is often followed by another.
Position sizing based on the ability to withstand continued declines is more important than guessing tomorrow.
Since 1999, $QQQ has dropped more than 5% in a single day a total of 58 times, with today's approximately −5.96% being the 58th; on the second trading day after the previous 57 times, only 49.1% rose, with an average of about +0.91% and a median of about 0%. $QQQ $SOXX $NVDA $SPY
These 58 days are highly clustered. The burst of the internet bubble from 2000 to 2001 accounts for a large portion: on 2000/1/4 it dropped 6.86%, then fell another 2.55% the next day; on 4/14 it dropped 8.66%, then rebounded 11.51% the next day. On 2001/9/17 it dropped 8.50%, some say it was after reopening post-9/11, and it fell another 2.24% the next day. During the 2008 financial crisis, on 9/29 it dropped 7.94% then rose 2.88% the next day; on 10/15 it dropped 8.96% then rose 5.42% the next day. On 2011/8/8 it dropped 6.02% then rose 4.82% the next day, some say it was due to US debt downgrade combined with European debt.
Recent dates include:
2020/3/9 −6.95%, next day +5.45%
2020/3/12 −9.17%, next day +8.47%
2020/3/16 −11.98%, next day +7.58%
2020/9/3 −5.07%, next day −1.33%
2022/5/5 −5.04%, next day −1.20%
2022/9/13 −5.48%, next day +0.79%
2025/4/3 −5.35%, next day −6.21%
2025/4/4 −6.21%, next day +0.24%
The pattern is simple: big drops cluster, and the next day is a coin toss—either a strong rebound or a sharp continuation down. The drop of 6.21% after 2025/4/3 is an example.
For today's drop, some say it is due to midterm election fears combined with AI and data center regulation panic plus chip sell-off. History only tells you the next day’s rise or fall is about half and half, but not the direction tomorrow.
After a big drop, leave it to discipline, not to gambling.
To put this round into history, first look at the three indices separately: the Philadelphia Semiconductor Index (SOX) has already reached the 20% level, while the Nasdaq and S&P 500 had not reached it by 11/4. $QQQ $SPY $SOXX $NVDA $AMD
On 11/4, SOX closed at 10268.65, down 3.13% for the day, a pullback of about 22% from around 13218 near 10/6, and it fell below 10447 again; from the high on 6/22 of about 14635 to the low on 7/29, it dropped about −28.6%. On the same day, the S&P 500 rose 1.6%. Some say $QQQ closed on 11/4 about 8.9% below the high on 10/6, not yet in a 20% bear market.
Some say that since around 1999, $QQQ-level drawdowns of about 20% or more have occurred at least six times: about −83% from 2000 to 2002, about −53% in 2008, about −23% in 2018, about −28.6% in 2020, about −35% to −36% in 2022, and about −23% from February to April 2025. Some say $SPY has had roughly four bear markets with drops over 20% since 2000: about −47.5% from 2000 to 2002, about −55% from 2007 to 2009, about −34% in 2020, and about −24.5% in 2022.
Some say since 1995, $SOX has had about 13 bear markets with drops over 20%, with a median depth of about −38%. The chip sector has long reached this level, while the index was still at a shallower position as of 11/4. Comparing apples to apples, don't mistake a sector bear market for an index crash.
Holding positions without moving and clearing out then re-entering—which is more costly depends on whether you can buy back in. Most people can't buy back in. $NVDA $SOXX
Clearing out then re-entering sounds reasonable: sell first, then buy back at a lower price after the drop. The problem is you have to get both points right. You need to sell at a high point and buy back at a low point, and have the courage to act when panic is at its peak. In reality, it's more common to sell halfway up the slope, then watch it rebound all the way, becoming increasingly hesitant to buy, and finally chasing it at a higher price.
Holding without moving also has a cost—you have to endure the drawdown. The semiconductor index has dropped about 22% from its high, and individual stocks often fall even more, causing significant psychological pressure. Watching your account every day is a drain.
My view depends on two things: position size and logic. If your position is so large that you can't sleep, reduce it to a size that lets you sleep. This has nothing to do with being bearish; it's about allowing yourself to think rationally and not be swayed by daily price fluctuations. If the logic still holds, you can keep the remaining position; if the logic breaks, exit completely without bargaining.
Whether the logic holds can be seen by the boundaries: cloud providers cutting capital expenditures, weakening DRAM and HBM, and the 10-year US Treasury yield returning above 5%. If these three conditions are intact, the only reason to clear out is emotional.
The last drop I mentioned before did not materialize, so I won't advise anyone to stubbornly hold a full position here. But selling everything due to one misjudgment and waiting for a perfect buy point is also a gamble.
Reducing positions is fine; clearing out requires a reason, and that reason can only come from the boundaries.
The day before yesterday, I was still happy about $AMD exceeding expectations
And today it's like this
The stock crash is like this
Is this the power of every leadership change?
So terrifying
What can be confirmed about this trend is that small-cap stocks, even star stocks like $SNDK, even if the trend returns, cannot reach their previous highs.
The semiconductor sector is really messed up this time, I don't know what to do.
Right now, I believe AI can't stop; they will eventually negotiate well.
But with this kind of situation, the growth itself is weak and the current circumstances,
basically, except for a few leading monopolies in hardware,
the rest are basically done for.
It's so sad
