|
13+ 年
经验
|
5
产品
|
697
演示版
|
|
0
工作
|
0
信号
|
0
订阅者
|
Sotirios Karkalakos
Gold Channel Rider is not one strategy. It is four, and each one has to pay for itself.
All four trade breakouts of gold's recent range, with a hard stop on every trade:
S1: 5-day channel, wide stop
S2: 10-day channel, trailing stop
S3: 20-day channel, trailing stop, near target
S4: 20-day channel, wide stop
MT5 Strategy Tester, 2013 - 2026, 0.5% risk per trade, each strategy on its own:
S1: 475 trades, profit factor 1.53, +5,634 USD
S2: 490 trades, profit factor 1.82, +10,717 USD
S3: 379 trades, profit factor 2.04, +9,531 USD
S4: 261 trades, profit factor 1.60, +3,901 USD
None of them is carried by the others. Different channel lengths catch different moves, so the losing stretches don't all arrive at once.
Each strategy has its own on/off switch. With a smaller account, switching off S1 and S4 (the wide-stop ones) lowers the risk of the minimum lot.
Free demo:
https://www.mql5.com/en/market/product/195647
Past results do not guarantee future results.
All four trade breakouts of gold's recent range, with a hard stop on every trade:
S1: 5-day channel, wide stop
S2: 10-day channel, trailing stop
S3: 20-day channel, trailing stop, near target
S4: 20-day channel, wide stop
MT5 Strategy Tester, 2013 - 2026, 0.5% risk per trade, each strategy on its own:
S1: 475 trades, profit factor 1.53, +5,634 USD
S2: 490 trades, profit factor 1.82, +10,717 USD
S3: 379 trades, profit factor 2.04, +9,531 USD
S4: 261 trades, profit factor 1.60, +3,901 USD
None of them is carried by the others. Different channel lengths catch different moves, so the losing stretches don't all arrive at once.
Each strategy has its own on/off switch. With a smaller account, switching off S1 and S4 (the wide-stop ones) lowers the risk of the minimum lot.
Free demo:
https://www.mql5.com/en/market/product/195647
Past results do not guarantee future results.
Sotirios Karkalakos
Don't trust EA screenshots. Including mine.
Every number I post about Gold Channel Rider comes from the MT5 Strategy Tester, and you can check it yourself in about three minutes with the free demo:
1. Symbol XAUUSD, period H1, model "1 minute OHLC" or "Every tick"
2. At least five years, for example 2020 to today. A trend follower judged on three months tells you almost nothing.
3. Deposit 10,000 USD or more, or load the "Small account" setfile for a smaller deposit.
What you should see (2013 - 2026, 0.5% risk): 13 of 14 years profitable. Some years are big (+29% in 2025, +24% in 2020), some are nearly flat (+1% in 2018), and 2022 was -5%. Flat stretches are part of trend following, and I'd rather show them than hide them.
If your result looks very different, write in the Comments tab and I will help you find why (usually the symbol name, the period or the deposit).
https://www.mql5.com/en/market/product/195647
Past results do not guarantee future results.
Every number I post about Gold Channel Rider comes from the MT5 Strategy Tester, and you can check it yourself in about three minutes with the free demo:
1. Symbol XAUUSD, period H1, model "1 minute OHLC" or "Every tick"
2. At least five years, for example 2020 to today. A trend follower judged on three months tells you almost nothing.
3. Deposit 10,000 USD or more, or load the "Small account" setfile for a smaller deposit.
What you should see (2013 - 2026, 0.5% risk): 13 of 14 years profitable. Some years are big (+29% in 2025, +24% in 2020), some are nearly flat (+1% in 2018), and 2022 was -5%. Flat stretches are part of trend following, and I'd rather show them than hide them.
If your result looks very different, write in the Comments tab and I will help you find why (usually the symbol name, the period or the deposit).
https://www.mql5.com/en/market/product/195647
Past results do not guarantee future results.
Sotirios Karkalakos
Passing a prop firm challenge is not about the good weeks. It is about surviving the bad one.
What Gold Channel Rider does to protect the account:
- Hard stop loss on every trade, sent with the order. Nothing is ever open without one.
- Flat for the weekend: no new orders late on Friday, everything closed before the market shuts. No Monday gap.
- Daily loss limit: reach it and the EA closes everything and pauses until the next day.
- Maximum drawdown stop, measured from peak equity: reach it and trading stops.
- No grid, no martingale, no averaging down.
There is a ready-made Prop firm setfile in the Comments tab of the product page: 0.25% risk per trade, 3% daily loss limit, 6% maximum drawdown. Load it, check your firm's own limits, and adjust if they are stricter.
Test it in the Strategy Tester first with the free demo:
https://www.mql5.com/en/market/product/195647
Past results do not guarantee future results.
What Gold Channel Rider does to protect the account:
- Hard stop loss on every trade, sent with the order. Nothing is ever open without one.
- Flat for the weekend: no new orders late on Friday, everything closed before the market shuts. No Monday gap.
- Daily loss limit: reach it and the EA closes everything and pauses until the next day.
- Maximum drawdown stop, measured from peak equity: reach it and trading stops.
- No grid, no martingale, no averaging down.
There is a ready-made Prop firm setfile in the Comments tab of the product page: 0.25% risk per trade, 3% daily loss limit, 6% maximum drawdown. Load it, check your firm's own limits, and adjust if they are stricter.
Test it in the Strategy Tester first with the free demo:
https://www.mql5.com/en/market/product/195647
Past results do not guarantee future results.
Sotirios Karkalakos
One input decides how aggressive Gold Channel Rider is: risk per trade.
Every trade is sized from its own stop loss, so "0.5% risk" means that if the stop is hit you lose 0.5% of the account. Not more.
Same rules, same 13.7 years of XAUUSD (MT5 Strategy Tester, 10,000 USD start):
0.25% risk -> 19,922 USD | 5.2% a year | max drawdown 8.5%
0.5% risk (default) -> 39,783 USD | 10.6% a year | max drawdown 9.3%
1% risk -> 164,808 USD | 22.7% a year | max drawdown 18.0%
The honest part: growth and drawdown rise together. There is no setting that gives you 1% returns with 0.25% pain. My advice is to start at 0.5% or lower.
One more honest detail: gold's stops are wide right now, so the minimum lot of 0.01 already risks about 75-150 USD per trade. For a true 0.5% plan on roughly 30,000 USD. Smaller account? Use the "Small account" setfile from the Comments tab.
https://www.mql5.com/en/market/product/195647
Past results do not guarantee future results.
Every trade is sized from its own stop loss, so "0.5% risk" means that if the stop is hit you lose 0.5% of the account. Not more.
Same rules, same 13.7 years of XAUUSD (MT5 Strategy Tester, 10,000 USD start):
0.25% risk -> 19,922 USD | 5.2% a year | max drawdown 8.5%
0.5% risk (default) -> 39,783 USD | 10.6% a year | max drawdown 9.3%
1% risk -> 164,808 USD | 22.7% a year | max drawdown 18.0%
The honest part: growth and drawdown rise together. There is no setting that gives you 1% returns with 0.25% pain. My advice is to start at 0.5% or lower.
One more honest detail: gold's stops are wide right now, so the minimum lot of 0.01 already risks about 75-150 USD per trade. For a true 0.5% plan on roughly 30,000 USD. Smaller account? Use the "Small account" setfile from the Comments tab.
https://www.mql5.com/en/market/product/195647
Past results do not guarantee future results.
Sotirios Karkalakos
A 92% win rate can still empty your account.
Before building Gold Channel Rider, I measured the mechanism most best-selling gold robots use: grid, recovery, no stop loss on any position, one equity stop. Four years of real XAUUSD data, with real spread, commission and swap.
156 winning series out of 169. A 92.3% win rate.
The account finished down 86.2%.
Both numbers are true at the same time. The 13 losing series were big enough to erase everything the 156 winners made, and more.
Gold Channel Rider wins 50.7% of its trades, and I show that number on purpose:
- every trade has a hard stop loss sent with the order
- losers are cut early, winners are left to run
- no grid, no martingale, no averaging down
Result over 13.7 years in the MT5 Strategy Tester: +298% at 0.5% risk per trade, max drawdown 9.3%.
Next time you see a 95% win rate, ask one question: where is the stop loss?
https://www.mql5.com/en/market/product/195647
Past results do not guarantee future results.
Before building Gold Channel Rider, I measured the mechanism most best-selling gold robots use: grid, recovery, no stop loss on any position, one equity stop. Four years of real XAUUSD data, with real spread, commission and swap.
156 winning series out of 169. A 92.3% win rate.
The account finished down 86.2%.
Both numbers are true at the same time. The 13 losing series were big enough to erase everything the 156 winners made, and more.
Gold Channel Rider wins 50.7% of its trades, and I show that number on purpose:
- every trade has a hard stop loss sent with the order
- losers are cut early, winners are left to run
- no grid, no martingale, no averaging down
Result over 13.7 years in the MT5 Strategy Tester: +298% at 0.5% risk per trade, max drawdown 9.3%.
Next time you see a 95% win rate, ask one question: where is the stop loss?
https://www.mql5.com/en/market/product/195647
Past results do not guarantee future results.
Sotirios Karkalakos
A 92% win rate can still empty your account.
Before building Gold Channel Rider, I measured the mechanism most best-selling gold robots use: grid, recovery, no stop loss on any position, one equity stop. Four years of real XAUUSD data, with real spread, commission and swap.
156 winning series out of 169. A 92.3% win rate.
The account finished down 86.2%.
Both numbers are true at the same time. The 13 losing series were big enough to erase everything the 156 winners made, and more.
Gold Channel Rider wins 50.7% of its trades, and I show that number on purpose:
- every trade has a hard stop loss sent with the order
- losers are cut early, winners are left to run
- no grid, no martingale, no averaging down
Result over 13.7 years in the MT5 Strategy Tester: +298% at 0.5% risk per trade, max drawdown 9.3%.
Next time you see a 95% win rate, ask one question: where is the stop loss?
https://www.mql5.com/en/market/product/195647
Past results do not guarantee future results.
Before building Gold Channel Rider, I measured the mechanism most best-selling gold robots use: grid, recovery, no stop loss on any position, one equity stop. Four years of real XAUUSD data, with real spread, commission and swap.
156 winning series out of 169. A 92.3% win rate.
The account finished down 86.2%.
Both numbers are true at the same time. The 13 losing series were big enough to erase everything the 156 winners made, and more.
Gold Channel Rider wins 50.7% of its trades, and I show that number on purpose:
- every trade has a hard stop loss sent with the order
- losers are cut early, winners are left to run
- no grid, no martingale, no averaging down
Result over 13.7 years in the MT5 Strategy Tester: +298% at 0.5% risk per trade, max drawdown 9.3%.
Next time you see a 95% win rate, ask one question: where is the stop loss?
https://www.mql5.com/en/market/product/195647
Past results do not guarantee future results.
Sotirios Karkalakos
A 92% win rate can still empty your account.
Before building Gold Channel Rider, I measured the mechanism most best-selling gold robots use: grid, recovery, no stop loss on any position, one equity stop. Four years of real XAUUSD data, with real spread, commission and swap.
156 winning series out of 169. A 92.3% win rate.
The account finished down 86.2%.
Both numbers are true at the same time. The 13 losing series were big enough to erase everything the 156 winners made, and more.
Gold Channel Rider wins 50.7% of its trades, and I show that number on purpose:
- every trade has a hard stop loss sent with the order
- losers are cut early, winners are left to run
- no grid, no martingale, no averaging down
Result over 13.7 years in the MT5 Strategy Tester: +298% at 0.5% risk per trade, max drawdown 9.3%.
Next time you see a 95% win rate, ask one question: where is the stop loss?
https://www.mql5.com/en/market/product/195647
Past results do not guarantee future results.
Before building Gold Channel Rider, I measured the mechanism most best-selling gold robots use: grid, recovery, no stop loss on any position, one equity stop. Four years of real XAUUSD data, with real spread, commission and swap.
156 winning series out of 169. A 92.3% win rate.
The account finished down 86.2%.
Both numbers are true at the same time. The 13 losing series were big enough to erase everything the 156 winners made, and more.
Gold Channel Rider wins 50.7% of its trades, and I show that number on purpose:
- every trade has a hard stop loss sent with the order
- losers are cut early, winners are left to run
- no grid, no martingale, no averaging down
Result over 13.7 years in the MT5 Strategy Tester: +298% at 0.5% risk per trade, max drawdown 9.3%.
Next time you see a 95% win rate, ask one question: where is the stop loss?
https://www.mql5.com/en/market/product/195647
Past results do not guarantee future results.
Sotirios Karkalakos
One rule made my gold EA better. And I chose it before I looked at the result.
Breakouts that go against gold's multi-month trend are the ones that fail. So Gold Channel Rider compares today's close with the close 120 days ago:
- above it: only buy-stops are placed
- below it: only sell-stops
Nothing else changes.
Same EA, 2013 - 2026, 0.5% risk per trade:
Profit factor: 1.37 -> 1.75
Max drawdown: 15.6% -> 9.3%
Losing years: 2 -> 1 (out of 14)
10,000 USD became 39,783 instead of 34,035
How I avoided curve fitting: I tested seven ideas on 2013-2020 only, then checked the survivors ONCE on 2021-2026. Only this filter held up out of sample. It also works on every lookback from 90 to 160 days, so 120 is not a lucky number.
Want the old behaviour? Set the trend filter to 0 and it trades both sides of every channel.
Free demo, test it on your own broker's data:
https://www.mql5.com/en/market/product/195647
Past results do not guarantee future results.
Breakouts that go against gold's multi-month trend are the ones that fail. So Gold Channel Rider compares today's close with the close 120 days ago:
- above it: only buy-stops are placed
- below it: only sell-stops
Nothing else changes.
Same EA, 2013 - 2026, 0.5% risk per trade:
Profit factor: 1.37 -> 1.75
Max drawdown: 15.6% -> 9.3%
Losing years: 2 -> 1 (out of 14)
10,000 USD became 39,783 instead of 34,035
How I avoided curve fitting: I tested seven ideas on 2013-2020 only, then checked the survivors ONCE on 2021-2026. Only this filter held up out of sample. It also works on every lookback from 90 to 160 days, so 120 is not a lucky number.
Want the old behaviour? Set the trend filter to 0 and it trades both sides of every channel.
Free demo, test it on your own broker's data:
https://www.mql5.com/en/market/product/195647
Past results do not guarantee future results.
Sotirios Karkalakos
One rule made my gold EA better. And I chose it before I looked at the result.
Breakouts that go against gold's multi-month trend are the ones that fail. So Gold Channel Rider compares today's close with the close 120 days ago:
- above it: only buy-stops are placed
- below it: only sell-stops
Nothing else changes.
Same EA, 2013 - 2026, 0.5% risk per trade:
Profit factor: 1.37 -> 1.75
Max drawdown: 15.6% -> 9.3%
Losing years: 2 -> 1 (out of 14)
10,000 USD became 39,783 instead of 34,035
How I avoided curve fitting: I tested seven ideas on 2013-2020 only, then checked the survivors ONCE on 2021-2026. Only this filter held up out of sample. It also works on every lookback from 90 to 160 days, so 120 is not a lucky number.
Want the old behaviour? Set the trend filter to 0 and it trades both sides of every channel.
Free demo, test it on your own broker's data:
https://www.mql5.com/en/market/product/195647
Past results do not guarantee future results.
Breakouts that go against gold's multi-month trend are the ones that fail. So Gold Channel Rider compares today's close with the close 120 days ago:
- above it: only buy-stops are placed
- below it: only sell-stops
Nothing else changes.
Same EA, 2013 - 2026, 0.5% risk per trade:
Profit factor: 1.37 -> 1.75
Max drawdown: 15.6% -> 9.3%
Losing years: 2 -> 1 (out of 14)
10,000 USD became 39,783 instead of 34,035
How I avoided curve fitting: I tested seven ideas on 2013-2020 only, then checked the survivors ONCE on 2021-2026. Only this filter held up out of sample. It also works on every lookback from 90 to 160 days, so 120 is not a lucky number.
Want the old behaviour? Set the trend filter to 0 and it trades both sides of every channel.
Free demo, test it on your own broker's data:
https://www.mql5.com/en/market/product/195647
Past results do not guarantee future results.
Sotirios Karkalakos
One rule made my gold EA better. And I chose it before I looked at the result.
Breakouts that go against gold's multi-month trend are the ones that fail. So Gold Channel Rider compares today's close with the close 120 days ago:
- above it: only buy-stops are placed
- below it: only sell-stops
Nothing else changes.
Same EA, 2013 - 2026, 0.5% risk per trade:
Profit factor: 1.37 -> 1.75
Max drawdown: 15.6% -> 9.3%
Losing years: 2 -> 1 (out of 14)
10,000 USD became 39,783 instead of 34,035
How I avoided curve fitting: I tested seven ideas on 2013-2020 only, then checked the survivors ONCE on 2021-2026. Only this filter held up out of sample. It also works on every lookback from 90 to 160 days, so 120 is not a lucky number.
Want the old behaviour? Set the trend filter to 0 and it trades both sides of every channel.
Free demo, test it on your own broker's data:
https://www.mql5.com/en/market/product/195647
Past results do not guarantee future results.
Breakouts that go against gold's multi-month trend are the ones that fail. So Gold Channel Rider compares today's close with the close 120 days ago:
- above it: only buy-stops are placed
- below it: only sell-stops
Nothing else changes.
Same EA, 2013 - 2026, 0.5% risk per trade:
Profit factor: 1.37 -> 1.75
Max drawdown: 15.6% -> 9.3%
Losing years: 2 -> 1 (out of 14)
10,000 USD became 39,783 instead of 34,035
How I avoided curve fitting: I tested seven ideas on 2013-2020 only, then checked the survivors ONCE on 2021-2026. Only this filter held up out of sample. It also works on every lookback from 90 to 160 days, so 120 is not a lucky number.
Want the old behaviour? Set the trend filter to 0 and it trades both sides of every channel.
Free demo, test it on your own broker's data:
https://www.mql5.com/en/market/product/195647
Past results do not guarantee future results.
Sotirios Karkalakos
Most gold EAs you see here show a 95% win rate and a perfectly straight equity line. Look closer and you will usually find a grid, a martingale, or no stop loss at all. Those curves are smooth right up to the day the account is gone.
Gold Channel Rider does the opposite.
- 4 breakout strategies on XAUUSD (5, 10 and 20-day channels)
- A hard stop loss on EVERY trade, sized by ATR
- No grid. No martingale. No averaging down. Never a bigger lot after a loss
- A 120-day trend filter: it only trades in the direction of gold's bigger move
- Friday close, spread and slippage guards, ready-made prop firm settings
MT5 Strategy Tester, Jan 2013 - Sep 2026, default settings, 0.5% risk per trade:
10,000 -> 39,783 USD
10.6% a year | max drawdown 9.3% | profit factor 1.75 | 1,605 trades
13 of 14 years profitable
The win rate is 50.7%, and I show it on purpose. A trend follower wins by letting winners run and cutting losers short, not by hiding losses.
Don't take my word for it. Download the free demo, run it in your own Strategy Tester with your own broker's data, and compare the result with the chart above. 4 setfiles (Default, Conservative, Small account, Prop firm) are in the Comments tab.
https://www.mql5.com/en/market/product/195647
Past results do not guarantee future results.
Gold Channel Rider does the opposite.
- 4 breakout strategies on XAUUSD (5, 10 and 20-day channels)
- A hard stop loss on EVERY trade, sized by ATR
- No grid. No martingale. No averaging down. Never a bigger lot after a loss
- A 120-day trend filter: it only trades in the direction of gold's bigger move
- Friday close, spread and slippage guards, ready-made prop firm settings
MT5 Strategy Tester, Jan 2013 - Sep 2026, default settings, 0.5% risk per trade:
10,000 -> 39,783 USD
10.6% a year | max drawdown 9.3% | profit factor 1.75 | 1,605 trades
13 of 14 years profitable
The win rate is 50.7%, and I show it on purpose. A trend follower wins by letting winners run and cutting losers short, not by hiding losses.
Don't take my word for it. Download the free demo, run it in your own Strategy Tester with your own broker's data, and compare the result with the chart above. 4 setfiles (Default, Conservative, Small account, Prop firm) are in the Comments tab.
https://www.mql5.com/en/market/product/195647
Past results do not guarantee future results.
Sotirios Karkalakos
已发布文章I built three gold grid EAs and measured them honestly
Every week someone posts a gold EA with a win rate above 95% and an equity curve that goes up like a staircase. I wanted to know whether that is a strategy or an accounting trick, so over the last month I built three grid systems for XAUUSD and tested each one until it told me the truth...
388
Sotirios Karkalakos
已发布产品
评论页附有四个现成的 set 文件：Default、Conservative、Small account（15,000 美元以下）和 Prop firm，并附有测试免费演示版时应使用的确切设置。 Gold Channel Rider 是一款用于 XAUUSD 的全自动智能交易系统（EA），基于系统化交易中最古老、并且在黄金上依然有效的思路之一：当价格突破多日区间时，往往会继续沿该方向运行。 它同时运行 四个相互独立的突破策略 ，分别基于 5 天、10 天和 20 天的价格通道。每笔交易都是挂在通道边缘的止损挂单， 每笔交易从开仓那一刻起就带有固定的止损和止盈 。 无网格 无马丁格尔 不摊平，不加仓回本 每个策略最多持有一个仓位 工作原理 EA 读取最近 120、240 或 480 根已收盘 H1 K 线的最高价和最低价。 在通道上方挂 buy stop，在下方挂 sell stop。一侧成交后，另一侧自动撤销。 止损和止盈按 ATR 计算，会随黄金当前的波动速度自动调整。 其中两个策略在盈利达到 1 个 ATR 后启动移动止损。 趋势过滤： 只在黄金最近 120
Sotirios Karkalakos
已发布产品
您的网格 EA 十笔里赢九笔。问题从来不在这里。 我把最畅销的黄金机器人所使用的机制——网格、加仓补救、任何持仓都没有止损、只有一个硬性的净值止损——放在四年真实 XAUUSD 数据上运行，点差取自每一根K线的真实记录，佣金和隔夜利息也都是真实的。 169 组序列中有 156 组盈利。胜率 92.3%。 账户最终亏损 86.2%，最大回撤 96.8%。 这两个数字同时为真，问题恰恰就在这里。92% 是写在销售页面上的那个；8% 是某个星期二找上门来的那个。 GRID GUARD 不替换您的 EA，它盯着您的 EA。 把它挂到您已经买下的那个机器人所在的同一张图表上。它读取已开仓位，从中测出网格的真实间距和手数递增比例，并回答您的 EA 从不告诉您的那一个问题： 如果它再加四层，这一篮持仓会让我付出多少？ 这个数字，以您的账户货币和净值百分比两种形式显示，字体大到隔着房间也能看清。 它怎么知道的 它从不查看您的 EA 的计划，也不需要。
·
1
Sotirios Karkalakos
已发布产品
为 MetaTrader 出售的东西几乎都在预测方向。我用十三年的 EURUSD 和黄金数据检验了这件事是否可能，做了前向滚动验证，并加了一次把标签随机打乱的对照运行。 方向的结果落在 48% 到 51% 之间。在黄金上，同样的模型在被打乱的标签上得到 53.4% —— 比在真实数据上还高。一个过拟合的系统从内部看就是这个样子，这也是回测曲线不能当作证据的原因。 明天会走多远，是另一个问题，而这个问题有答案。VolGuard 建立在这个答案之上，除此之外别无其他。 它做什么 每天日线收盘时，VolGuard 预测市场明天会走多远，并把这个预测变成一个数字：你的仓位大小应该乘以的系数。 预计走幅是平常两倍的一天，只给一半的仓位，这样承担的资金留在你放它的地方，而不是悄悄翻倍。平静的一天则多给一些。超过你自己设定的上限时，它会说今天不要交易 —— 因为在剧烈的一天里，小仓位仍然是仓位。 系数在两端都被刻意限制。不限制的话，一个非常平静的预测会发出五倍杠杆，而平静预测出错的那一天，恰恰就是要命的那一天。 面板 状态只用一个词：NORMAL、REDUCED、BLOCKED 或
Sotirios Karkalakos
已发布产品
Most traders who fail a prop firm challenge do not fail because they did not know the rules. They fail because, at the moment it mattered, nobody was doing the arithmetic. Prop Guard Monitor does the arithmetic. It sits on one chart, watches the whole account, and answers one question continuously: how much room is left before a rule breaks. It never closes a position. It never deletes an order. It never locks anything. This build cannot touch a trade even if you want it to - the trading
Sotirios Karkalakos
已发布产品
丢掉一个资金账户，往往不是因为那笔让你后悔的交易，而是因为日内亏损上限被越过时、仍然开着的那一笔。 Prop Guard 挂在一张图表上，监控整个账户。它不开仓，也不发信号。它只持续回答一个问题：距离触发规则还剩多少空间——当答案足够接近零时，它就动手。 大多数防护工具做错的地方 回撤峰值会写入磁盘。关闭终端、重启电脑、一周后再连接：你的峰值仍然是你的峰值。只把峰值放在内存里的工具，要求你每天早上重新挂载一次；而一个会忘记峰值的追踪回撤规则，根本不是追踪回撤规则，只是一个长得像它的数字。 它监控什么 日内亏损。百分比可以按初始余额计算，也可以按当天开始时的余额计算；亏损可以从当天的起始余额衡量，也可以从当天的起始净值衡量。各家公司的表述不同，而这个差别就是真金白银，所以两者都是可设置项，而不是假设。 最大回撤。可选按初始余额的静态回撤，或按账户曾达到的最高余额、最高净值的追踪回撤。